gl_context.cxx

#include <cgv/base/group.h>
#include "gl_context.h"
#include "gl_tools.h"
#include <cgv_gl/gl/wgl.h>
#ifdef _WIN32
#undef TA_LEFT
#undef TA_TOP
#undef TA_RIGHT
#undef TA_BOTTOM
#endif
#include <cgv/base/base.h>
#include <cgv/base/action.h>
#include <cgv/render/drawable.h>
#include <cgv/render/shader_program.h>
#include <cgv/render/attribute_array_binding.h>
#include <cgv/render/vertex_buffer.h>
#include <cgv/render/textured_material.h>
#include <cgv/utils/scan.h>
#include <cgv/media/image/image_writer.h>
#include <cgv/gui/event_handler.h>
#include <cgv/math/ftransform.h>
#include <cgv/math/geom.h>
#include <cgv/math/inv.h>
#include <cgv/type/standard_types.h>
#include <cgv/os/clipboard.h>
 
using namespace cgv::base;
using namespace cgv::type;
using namespace cgv::type::info;
using namespace cgv::gui;
using namespace cgv::os;
using namespace cgv::math;
using namespace cgv::media::font;
using namespace cgv::media::illum;
 
namespace cgv {
    namespace render {
        namespace gl {
 
std::vector<int> get_context_creation_attrib_list(cgv::render::context_config& cc)
{
#ifdef _WIN32
    const int CONTEXT_FLAGS = WGL_CONTEXT_FLAGS_ARB;
    const int CONTEXT_FORWARD_COMPATIBLE_BIT = WGL_CONTEXT_FORWARD_COMPATIBLE_BIT_ARB;
    const int CONTEXT_DEBUG_BIT = WGL_CONTEXT_DEBUG_BIT_ARB;
    const int CONTEXT_MAJOR_VERSION = WGL_CONTEXT_MAJOR_VERSION_ARB;
    const int CONTEXT_MINOR_VERSION = WGL_CONTEXT_MINOR_VERSION_ARB;
    const int CONTEXT_PROFILE_MASK = WGL_CONTEXT_PROFILE_MASK_ARB;
    const int CONTEXT_CORE_PROFILE_BIT = WGL_CONTEXT_CORE_PROFILE_BIT_ARB;
    const int CONTEXT_COMPATIBILITY_PROFILE_BIT = WGL_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB;
#else
    const int CONTEXT_FLAGS = GLX_CONTEXT_FLAGS_ARB;
    const int CONTEXT_FORWARD_COMPATIBLE_BIT = GLX_CONTEXT_FORWARD_COMPATIBLE_BIT_ARB;
    const int CONTEXT_DEBUG_BIT = GLX_CONTEXT_DEBUG_BIT_ARB;
    const int CONTEXT_MAJOR_VERSION = GLX_CONTEXT_MAJOR_VERSION_ARB;
    const int CONTEXT_MINOR_VERSION = GLX_CONTEXT_MINOR_VERSION_ARB;
    const int CONTEXT_PROFILE_MASK = GLX_CONTEXT_PROFILE_MASK_ARB;
    const int CONTEXT_CORE_PROFILE_BIT = GLX_CONTEXT_CORE_PROFILE_BIT_ARB;
    const int CONTEXT_COMPATIBILITY_PROFILE_BIT = GLX_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB;
#endif
    std::vector<int> attrib_list;
    int version = 0;
    if (cc.version_major < 1)
        version = 46;
    else
        version = 10 * cc.version_major + (cc.version_minor >= 0 ? cc.version_minor : 0);
    if ((version > 30 && cc.forward_compatible) || cc.debug) {
        attrib_list.push_back(CONTEXT_FLAGS);
        attrib_list.push_back(
            (cc.forward_compatible ? CONTEXT_FORWARD_COMPATIBLE_BIT : 0) +
            (cc.debug ? CONTEXT_DEBUG_BIT : 0));
    }
    if (cc.version_major > 0) {
        attrib_list.push_back(CONTEXT_MAJOR_VERSION);
        attrib_list.push_back(cc.version_major);
    }
    if (cc.version_minor >= 0) {
        attrib_list.push_back(CONTEXT_MINOR_VERSION);
        attrib_list.push_back(cc.version_minor);
    }
    if (version > 31) {
        attrib_list.push_back(CONTEXT_PROFILE_MASK);
        attrib_list.push_back(
            cc.core_profile ? CONTEXT_CORE_PROFILE_BIT : CONTEXT_COMPATIBILITY_PROFILE_BIT);
    }
    attrib_list.push_back(0);
    return attrib_list;
}
 
GLenum map_to_gl(PrimitiveType primitive_type)
{
    static const GLenum gl_primitive_type[] = {
        GLenum(-1),
        GL_POINTS,
        GL_LINES,
        GL_LINES_ADJACENCY,
        GL_LINE_STRIP,
        GL_LINE_STRIP_ADJACENCY,
        GL_LINE_LOOP,
        GL_TRIANGLES,
        GL_TRIANGLES_ADJACENCY,
        GL_TRIANGLE_STRIP,
        GL_TRIANGLE_STRIP_ADJACENCY,
        GL_TRIANGLE_FAN,
        GL_QUADS,
        GL_QUAD_STRIP,
        GL_POLYGON,
        GL_PATCHES
    };
    return gl_primitive_type[primitive_type];
}
 
GLenum map_to_gl(MaterialSide material_side)
{
    static const GLenum gl_material_side[] = {
        GLenum(0),
        GL_FRONT,
        GL_BACK,
        GL_FRONT_AND_BACK
    };
    return gl_material_side[material_side];
}
 
GLenum map_to_gl(AccessType access_type)
{
    static const GLenum gl_access_type[] = {
        GL_READ_ONLY,
        GL_WRITE_ONLY,
        GL_READ_WRITE
    };
    return gl_access_type[access_type];
}
 
GLenum map_to_gl(BlendFunction blend_function)
{
    static const GLenum gl_blend_func[] = {
        GL_ZERO,
        GL_ONE,
        GL_SRC_COLOR,
        GL_ONE_MINUS_SRC_COLOR,
        GL_DST_COLOR,
        GL_ONE_MINUS_DST_COLOR,
        GL_SRC_ALPHA,
        GL_ONE_MINUS_SRC_ALPHA,
        GL_DST_ALPHA,
        GL_ONE_MINUS_DST_ALPHA,
        GL_CONSTANT_COLOR,
        GL_ONE_MINUS_CONSTANT_COLOR,
        GL_CONSTANT_ALPHA,
        GL_ONE_MINUS_CONSTANT_ALPHA,
        GL_SRC_ALPHA_SATURATE,
        GL_SRC1_COLOR,
        GL_ONE_MINUS_SRC1_COLOR,
        GL_SRC1_ALPHA,
        GL_ONE_MINUS_SRC1_ALPHA
    };
    return gl_blend_func[blend_function];
}
 
GLenum map_to_gl(CompareFunction compare_func)
{
    static const GLenum gl_compare_func[] = {
        GL_LEQUAL,
        GL_GEQUAL,
        GL_LESS,
        GL_GREATER,
        GL_EQUAL,
        GL_NOTEQUAL,
        GL_ALWAYS,
        GL_NEVER
    };
    return gl_compare_func[compare_func];
}
 
static const GLenum gl_depth_format_ids[] =
{
    GL_DEPTH_COMPONENT,
    GL_DEPTH_COMPONENT16,
    GL_DEPTH_COMPONENT24,
    GL_DEPTH_COMPONENT32
};
 
static const GLenum gl_color_buffer_format_ids[] =
{
    GL_RGB,
    GL_RGBA
};
 
static const char* depth_formats[] =
{
    "[D]",
    "uint16[D]",
    "uint32[D:24]",
    "uint32[D]",
    0
};
 
static const char* color_buffer_formats[] =
{
    "[R,G,B]",
    "[R,G,B,A]",
    0
};
 
GLenum get_tex_dim(TextureType texture_type)
{
    static const GLenum gl_texture_type[] = {
        GLenum(0),
        GL_TEXTURE_1D,
        GL_TEXTURE_2D,
        GL_TEXTURE_3D,
        GL_TEXTURE_1D_ARRAY,
        GL_TEXTURE_2D_ARRAY,
        GL_TEXTURE_CUBE_MAP,
        GL_TEXTURE_2D_MULTISAMPLE,
        GL_TEXTURE_2D_MULTISAMPLE_ARRAY
    };
    return gl_texture_type[texture_type];
}
 
GLenum get_tex_bind(TextureType texture_type)
{
    static const GLenum gl_tex_binding[] = {
        GLenum(0),
        GL_TEXTURE_BINDING_1D,
        GL_TEXTURE_BINDING_2D,
        GL_TEXTURE_BINDING_3D,
        GL_TEXTURE_BINDING_1D_ARRAY,
        GL_TEXTURE_BINDING_2D_ARRAY,
        GL_TEXTURE_BINDING_CUBE_MAP,
        GL_TEXTURE_BUFFER,
        GL_TEXTURE_BINDING_2D_MULTISAMPLE,
        GL_TEXTURE_BINDING_2D_MULTISAMPLE_ARRAY
    };
    return gl_tex_binding[texture_type];
}
 
GLenum map_to_gl(TextureWrap texture_wrap)
{
    static const GLenum gl_texture_wrap[] = {
        GL_REPEAT,
        GL_CLAMP,
        GL_CLAMP_TO_EDGE,
        GL_CLAMP_TO_BORDER,
        GL_MIRROR_CLAMP_EXT,
        GL_MIRROR_CLAMP_TO_EDGE_EXT,
        GL_MIRROR_CLAMP_TO_BORDER_EXT,
        GL_MIRRORED_REPEAT
    };
    return gl_texture_wrap[texture_wrap];
}
 
GLenum map_to_gl(TextureFilter filter_type)
{
    static const GLenum gl_texture_filter[] = {
        GL_NEAREST,
        GL_LINEAR,
        GL_NEAREST_MIPMAP_NEAREST,
        GL_LINEAR_MIPMAP_NEAREST,
        GL_NEAREST_MIPMAP_LINEAR,
        GL_LINEAR_MIPMAP_LINEAR,
        GL_LINEAR_MIPMAP_LINEAR
    };
    return gl_texture_filter[filter_type];
}
 
GLboolean map_to_gl(bool flag) {
    return flag ? GL_TRUE : GL_FALSE;
}
 
GLuint get_gl_id(const void* handle)
{
    return (const GLuint&)handle - 1;
}
 
void* get_handle(GLuint id)
{
    void* handle = 0;
    (GLuint&)handle = id + 1;
    return handle;
}
 
void gl_context::put_id(void* handle, void* ptr) const
{
    *static_cast<GLuint*>(ptr) = get_gl_id(handle);
}

void set_gl_format(texture& tex, GLuint gl_format, const std::string& component_format_description)
{
    tex.set_component_format(component_format_description);
    (GLuint&)tex.internal_format = gl_format;
}

GLuint get_gl_format(const texture& tex)
{
    if (tex.internal_format)
        return (const GLuint&)tex.internal_format;
    cgv::data::component_format best_cf;
    GLuint gl_format = find_best_texture_format(tex, &best_cf);
    return gl_format;
}
 
 
void gl_set_material_color(GLenum side, const cgv::media::illum::phong_material::color_type& c, float alpha, GLenum type)
{
    GLfloat v[4] = { c[0],c[1],c[2],c[3] * alpha };
    glMaterialfv(side, type, v);
}

void gl_set_material(const cgv::media::illum::phong_material& mat, MaterialSide ms, float alpha)
{
    if (ms == MS_NONE)
        return;
    unsigned side = map_to_gl(ms);
    gl_set_material_color(side, mat.ambient, alpha, GL_AMBIENT);
    gl_set_material_color(side, mat.diffuse, alpha, GL_DIFFUSE);
    gl_set_material_color(side, mat.specular, alpha, GL_SPECULAR);
    gl_set_material_color(side, mat.emission, alpha, GL_EMISSION);
    glMaterialf(side, GL_SHININESS, mat.shininess);
}

float gl_context::get_info_font_size() const
{
    return info_font_size;
}

cgv::media::font::font_face_ptr gl_context::get_info_font_face() const
{
    if (info_font_face.empty()) {
        font_ptr info_font = default_font(true);
        if (!info_font.empty())
            info_font_face = info_font->get_font_face(FFA_REGULAR);
    }
    return info_font_face;
}

gl_context::gl_context()
{
    frame_buffer_stack.top()->handle = get_handle(0);
    max_nr_indices = 0;
    max_nr_vertices = 0;
    info_font_size = 14;
    show_help = false;
    show_stats = false;
 
    // set initial GL state from stack contents
    set_bg_color(get_bg_color());
    set_bg_depth(get_bg_depth());
    set_bg_stencil(get_bg_stencil());
    set_bg_accum_color(get_bg_accum_color());
 
    set_depth_test_state(get_depth_test_state());
    set_cull_state(get_cull_state());
    set_blend_state(get_blend_state());
    set_buffer_mask(get_buffer_mask());
}

RenderAPI gl_context::get_render_api() const
{
    return RA_OPENGL;
}
 
//GL_STACK_OVERFLOW, "stack overflow"
//GL_STACK_UNDERFLOW, "stack underflow",
std::string get_source_tag_name(GLenum tag)
{
    static std::map<GLenum, const char*> source_tags = {
            { GL_DEBUG_SOURCE_API,             "api" },
            { GL_DEBUG_SOURCE_WINDOW_SYSTEM,   "window system" },
            { GL_DEBUG_SOURCE_SHADER_COMPILER, "shader compiler" },
            { GL_DEBUG_SOURCE_THIRD_PARTY,     "3rd party" },
            { GL_DEBUG_SOURCE_APPLICATION,     "application" },
            { GL_DEBUG_SOURCE_OTHER,           "other" }
    };
    return source_tags[tag];
}
 
std::string get_type_tag_name(GLenum tag)
{
    static std::map<GLenum, const char*> type_tags = {
            { GL_DEBUG_TYPE_ERROR,               "error" },
            { GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR, "deprecated behavior" },
            { GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR,  "undefinded behavior" },
            { GL_DEBUG_TYPE_PORTABILITY,         "portability" },
            { GL_DEBUG_TYPE_PERFORMANCE,         "performance" },
            { GL_DEBUG_TYPE_OTHER,               "other" },
            { GL_DEBUG_TYPE_MARKER,              "marker" },
            { GL_DEBUG_TYPE_PUSH_GROUP,          "push group" },
            { GL_DEBUG_TYPE_POP_GROUP,           "pop group" }
    };
    return type_tags[tag];
}
 
std::string get_severity_tag_name(GLenum tag)
{
    static std::map<GLenum, const char*> severity_tags = {
            { GL_DEBUG_SEVERITY_NOTIFICATION, "notification" },
            { GL_DEBUG_SEVERITY_HIGH,         "high" },
            { GL_DEBUG_SEVERITY_MEDIUM,       "medium" },
            { GL_DEBUG_SEVERITY_LOW,          "low" }
    };
    return severity_tags[tag];
};
 
void GLAPIENTRY debug_callback(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar* message, const void* userParam)
{
    if (severity == GL_DEBUG_SEVERITY_NOTIFICATION)
        return;
    const gl_context* ctx = reinterpret_cast<const gl_context*>(userParam);
    std::string msg(message, length);
    msg = std::string("GLDebug Message[") + cgv::utils::to_string(id) + "] from " + get_source_tag_name(source) + " of type " + get_type_tag_name(type) + " of severity " + get_severity_tag_name(severity) + "\n" + msg;
    ctx->error(msg);
}

bool gl_context::configure_gl()
{
    if (!ensure_glew_initialized()) {
        error("gl_context::configure_gl could not initialize glew");
        return false;
    }
    const GLubyte* version_string = glGetString(GL_VERSION);
    version_major = version_string[0] - '0';
    version_minor = version_string[2] - '0';
    if (version_major >= 3) {
        GLint context_flags;
        glGetIntegerv(GL_CONTEXT_FLAGS, &context_flags);
        // weird behavior under windows or just nvidia or just my laptop (Stefan)??
        debug = (context_flags & GL_CONTEXT_FLAG_DEBUG_BIT) != 0;
        forward_compatible = (context_flags & GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT) != 0;
    }
    else {
        debug = false;
        forward_compatible = false;
    }
    int version = 10 * version_major + version_minor;
    if (version >= 32) {
        GLint context_profile;
        glGetIntegerv(GL_CONTEXT_PROFILE_MASK, &context_profile);
#ifdef _DEBUG
        if (context_profile != 0) {
            if (core_profile != ((context_profile & GL_CONTEXT_COMPATIBILITY_PROFILE_BIT) == 0)) {
                if (core_profile)
                    std::cerr << "WARNING: Ignoring that GL_CONTEXT_PROFILE_MASK yields compatibility profile for context created as core." << std::endl;
                else
                    std::cerr << "WARNING: Ignoring that GL_CONTEXT_PROFILE_MASK does not yield compatibility profile for context created as such." << std::endl;
            }
        }
        else {
            std::cerr << "WARNING: Ignoring that GL_CONTEXT_PROFILE_MASK yields zero." << std::endl;
        }
#endif
    }
    else
        core_profile = false;
    const GLubyte* vendor_c_string = glGetString(GL_VENDOR);
    std::string vendor_string(reinterpret_cast<const char*>(vendor_c_string));
    vendor_string = cgv::utils::to_upper(vendor_string);
    
    if (vendor_string.find("NVIDIA") != std::string::npos)
        gpu_vendor = GPU_VENDOR_NVIDIA;
    else if (vendor_string.find("INTEL") != std::string::npos)
        gpu_vendor = GPU_VENDOR_INTEL;
    else if (vendor_string.find("AMD") != std::string::npos || vendor_string.find("ATI") != std::string::npos)
        gpu_vendor = GPU_VENDOR_AMD;
    
    // query device capabilities
    glGetIntegerv(GL_MAX_RENDERBUFFER_SIZE, &gpu_capabilities.max_render_buffer_size);
    glGetIntegerv(GL_MAX_GEOMETRY_OUTPUT_VERTICES, &gpu_capabilities.max_geometry_shader_output_vertex_count);
    glGetIntegerv(GL_MAX_GEOMETRY_OUTPUT_COMPONENTS, &gpu_capabilities.max_geometry_shader_output_component_count);
    glGetIntegerv(GL_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS, &gpu_capabilities.max_geometry_shader_total_output_component_count);
    glGetIntegerv(GL_MAX_COMPUTE_SHARED_MEMORY_SIZE, &gpu_capabilities.max_compute_shared_memory_size);
    glGetIntegerv(GL_MAX_COMPUTE_WORK_GROUP_INVOCATIONS, &gpu_capabilities.max_compute_work_group_invocations);
    for(unsigned i = 0; i < 3; ++i)
        glGetIntegeri_v(GL_MAX_COMPUTE_WORK_GROUP_COUNT, static_cast<GLuint>(i), &gpu_capabilities.max_compute_work_group_count[i]);
    for(unsigned i = 0; i < 3; ++i)
        glGetIntegeri_v(GL_MAX_COMPUTE_WORK_GROUP_SIZE, static_cast<GLuint>(i), &gpu_capabilities.max_compute_work_group_size[i]);
    
#ifdef _DEBUG
    std::cout << "OpenGL version " << version_major << "." << version_minor << (core_profile?" (core)":" (compatibility)") << (debug?" (debug)":"") << (forward_compatible?" (forward_compatible)":"") << std::endl;
    const GLubyte* renderer_c_string = glGetString(GL_RENDERER);
    const GLubyte* glslversion_c_string = glGetString(GL_SHADING_LANGUAGE_VERSION);
    if (vendor_c_string)
        std::cout << "   vendor     : " << vendor_c_string << std::endl;
    if (renderer_c_string)
        std::cout << "   renderer   : " << renderer_c_string << std::endl;
    if (glslversion_c_string)
        std::cout << "   glslversion: " << glslversion_c_string << std::endl;
#endif
    if (debug) {
        glEnable(GL_DEBUG_OUTPUT);
        if (version >= 43) {
            glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS);
            if (!check_gl_error("gl_context::configure() debug output"))
                glDebugMessageCallback(debug_callback, this);
        }
    }
    // use the eye location to compute the specular lighting
    if (!core_profile) {
        glLightModelf(GL_LIGHT_MODEL_LOCAL_VIEWER, 0);
        // this makes opengl normalize all surface normals before lighting calculations,
        // which is essential when using scaling to deform tesselated primities
        glEnable(GL_NORMALIZE);
 
        // should be initialized by the driver, but better be safe than risk errors later
        glMatrixMode(GL_MODELVIEW);
    }
    set_viewport(ivec4(0, 0, get_width(), get_height()));
//  if (check_gl_error("gl_context::configure_gl before init of children"))
//      return false;
    
    group_ptr grp(dynamic_cast<group*>(this));
    single_method_action<cgv::render::drawable, bool, cgv::render::context&> sma(*this, &drawable::init, false, false);
    for (unsigned i = 0; i<grp->get_nr_children(); ++i)
        traverser(sma, "nc").traverse(grp->get_child(i));
 
//  if (check_gl_error("gl_context::configure_gl after init of children."))
//      return false;
 
    return true;
}
 
void gl_context::resize_gl()
{
    group_ptr grp(dynamic_cast<group*>(this));
    set_viewport(ivec4(0, 0, get_width(), get_height()));
    if (grp) {
        single_method_action_2<drawable, void, unsigned int, unsigned int> sma(get_width(), get_height(), &drawable::resize);
        traverser(sma).traverse(grp);
    }
}
 
void gl_context::set_bg_color(vec4 rgba) {
    glClearColor(rgba[0], rgba[1], rgba[2], rgba[3]);
    context::set_bg_color(rgba);
}
 
void gl_context::set_bg_depth(float d) {
    glClearDepth(d);
    context::set_bg_depth(d);
}
 
void gl_context::set_bg_stencil(int s) {
    glClearStencil(s);
    context::set_bg_stencil(s);
}
 
void gl_context::set_bg_accum_color(vec4 rgba) {
    if(!core_profile)
        glClearAccum(rgba[0], rgba[1], rgba[2], rgba[3]);
    context::set_bg_accum_color(rgba);
}
 
void gl_context::clear_background(bool color_flag, bool depth_flag, bool stencil_flag, bool accum_flag) {
    GLenum bits = 0;
    if(color_flag)
        bits |= GL_COLOR_BUFFER_BIT;
    if(depth_flag)
        bits |= GL_DEPTH_BUFFER_BIT;
    if(stencil_flag)
        bits |= GL_STENCIL_BUFFER_BIT;
    if(accum_flag && !core_profile)
        bits |= GL_ACCUM_BUFFER_BIT;
    if(bits)
        glClear(bits);
}

float gl_context::get_current_font_size() const
{
    if (current_font_size == 0)
        return get_info_font_size();
    return current_font_size;
}

media::font::font_face_ptr gl_context::get_current_font_face() const
{
    if (current_font_face.empty())
        return get_info_font_face();
    return current_font_face;
}
 
void gl_context::init_render_pass()
{
#ifdef WIN32
    wglSwapIntervalEXT(enable_vsync ? 1 : 0);
#else
    glXSwapIntervalEXT(glXGetCurrentDisplay(), glXGetCurrentDrawable(), enable_vsync ? 1 : 0);
#endif
    if (sRGB_framebuffer)
        glEnable(GL_FRAMEBUFFER_SRGB);
    else
        glDisable(GL_FRAMEBUFFER_SRGB);
 
    static cgv::render::RenderPassFlags last_render_pass_flags = get_default_render_pass_flags();
    cgv::render::RenderPassFlags current_render_pass_flags = get_render_pass_flags();
    if (current_render_pass_flags & RPF_SET_LIGHTS) {
        for (unsigned i = 0; i < nr_default_light_sources; ++i)
            set_light_source(default_light_source_handles[i], default_light_source[i], false);
 
        for (unsigned i = 0; i < nr_default_light_sources; ++i)
            if (current_render_pass_flags & RPF_SET_LIGHTS_ON)
                enable_light_source(default_light_source_handles[i]);
            else
                disable_light_source(default_light_source_handles[i]);
    }
    else if ((last_render_pass_flags & RPF_SET_LIGHTS) == 0) {
        for (unsigned i = 0; i < nr_default_light_sources; ++i)
            if (is_light_source_enabled(default_light_source_handles[i]))
                disable_light_source(default_light_source_handles[i]);
    }
    last_render_pass_flags = current_render_pass_flags;
 
    if (get_render_pass_flags()&RPF_SET_MATERIAL) {
        set_material(default_material);
    }
    if ((get_render_pass_flags()&RPF_ENABLE_MATERIAL) && !core_profile) {
        // this mode allows to define the ambient and diffuse color of the surface material
        // via the glColor commands
        glEnable(GL_COLOR_MATERIAL);
    }
    if (get_render_pass_flags()&RPF_SET_STATE_FLAGS) {
        // set some default settings
        enable_depth_test();
        set_cull_state(CM_BACKFACE);
        if (!core_profile)
            glEnable(GL_NORMALIZE);
    }
    if ((get_render_pass_flags()&RPF_SET_PROJECTION) != 0)
        set_projection_matrix(cgv::math::perspective4<double>(45.0, (double)get_width()/get_height(),0.001,1000.0));
 
    if ((get_render_pass_flags()&RPF_SET_MODELVIEW) != 0)
        set_modelview_matrix(cgv::math::look_at4<double>(vec3(0,0,10), vec3(0,0,0), vec3(0,1,0)));
    
    if (check_gl_error("gl_context::init_render_pass before init_frame"))
        return;
 
    group* grp = dynamic_cast<group*>(this);
    if (grp && (get_render_pass_flags()&RPF_DRAWABLES_INIT_FRAME)) {
        single_method_action<drawable,void,cgv::render::context&> sma(*this, &drawable::init_frame, true, true);
        traverser(sma).traverse(group_ptr(grp));
    }
 
    if (check_gl_error("gl_context::init_render_pass after init_frame"))
        return;
    // this defines the background color to which the frame buffer is set by glClear
    if(get_render_pass_flags() & RPF_SET_CLEAR_COLOR)
        set_bg_color(get_bg_color());
    // this defines the background depth buffer value set by glClear
    if(get_render_pass_flags() & RPF_SET_CLEAR_DEPTH)
        set_bg_depth(get_bg_depth());
    // this defines the background depth buffer value set by glClear
    if(get_render_pass_flags() & RPF_SET_CLEAR_STENCIL)
        set_bg_stencil(get_bg_stencil());
    // this defines the background color to which the accum buffer is set by glClear
    if(get_render_pass_flags() & RPF_SET_CLEAR_ACCUM && !core_profile)
        set_bg_accum_color(get_bg_accum_color());
    clear_background(
        get_render_pass_flags() & RPF_CLEAR_COLOR,
        get_render_pass_flags() & RPF_CLEAR_DEPTH,
        get_render_pass_flags() & RPF_CLEAR_STENCIL,
        get_render_pass_flags() & RPF_CLEAR_ACCUM
    );
}

void gl_context::finish_render_pass()
{
}
 
struct format_callback_handler : public traverse_callback_handler
{
    std::ostream& os;
    format_callback_handler(std::ostream& _os) : os(_os) 
    {
    }
    bool on_enter_children(group_ptr) override
    {
        os << "\a";
        return false;
    }

    bool on_leave_children(group_ptr) override
    {
        os << "\b";
        return false;
    }
 
};
 
void gl_context::draw_textual_info()
{
    if (show_help || show_stats) {
        rgba tmp = current_color;
        push_depth_test_state();
        disable_depth_test();
 
        push_pixel_coords();
        enable_font_face(get_info_font_face(), get_info_font_size());
 
        set_cursor(20, get_height()-1-20);
 
        vec4 bg = get_bg_color();
        //if (bg_r + bg_g + bg_b < 1.5f)
        if (bg[0] + bg[1] + bg[2] < 1.5f)
            set_color(rgba(1, 1, 1, 1));
        else
            set_color(rgba(0, 0, 0, 1));
 
        // traverse objects for show_stats callback
        format_callback_handler fch(output_stream());
        group_ptr grp(dynamic_cast<group*>(this));
        if (grp && show_stats) {
            single_method_action<cgv::base::base, void, std::ostream&> sma(output_stream(), &cgv::base::base::stream_stats, false, false);
            traverser(sma, "nc").traverse(grp, &fch);
            output_stream() << std::endl;
        }
        //if (bg_r + bg_g + bg_b < 1.5f)
        if(bg[0] + bg[1] + bg[2] < 1.5f)
            set_color(rgba(1, 1, 0, 1));
        else
            set_color(rgba(0.4f, 0.3f, 0, 1));
 
        if (grp && show_help) {
            // collect help from myself and all children
            single_method_action<event_handler, void, std::ostream&> sma(output_stream(), &event_handler::stream_help, false, false);
            traverser(sma, "nc").traverse(grp, &fch);
            output_stream().flush();
        }
        pop_pixel_coords();
        pop_depth_test_state();
    }
}
 
void gl_context::perform_screen_shot()
{
    glFlush();
    data::data_view dv;
    if (!read_frame_buffer(dv))
        return;
    std::string ext("bmp");
    std::string exts = cgv::media::image::image_writer::get_supported_extensions();
    if (cgv::utils::is_element("png",exts))
        ext = "png";
    else if (cgv::utils::is_element("tif",exts))
        ext = "tif";
    cgv::media::image::image_writer wr(std::string("screen_shot.")+ext);
    if (wr.is_format_supported(*dv.get_format()))
        wr.write_image(dv);
}

void gl_context::enumerate_program_uniforms(shader_program& prog, std::vector<std::string>& names, std::vector<int>* locations_ptr, std::vector<int>* sizes_ptr, std::vector<int>* types_ptr, bool show) const
{
    GLint count;
    glGetProgramiv(get_gl_id(prog.handle), GL_ACTIVE_UNIFORMS, &count);
    for (int i = 0; i < count; ++i) {
        GLchar name[1000];
        GLsizei length;
        GLint size;
        GLenum type;
        glGetActiveUniform(get_gl_id(prog.handle), i, 1000, &length, &size, &type, name);
        std::string name_str(name, length);
        names.push_back(name_str);
        if (sizes_ptr)
            sizes_ptr->push_back(size);
        if (types_ptr)
            types_ptr->push_back(type);
        int loc = glGetUniformLocation(get_gl_id(prog.handle), name_str.c_str());
        if (locations_ptr)
            locations_ptr->push_back(loc);
        if (show)
            std::cout << i << " at " << loc << " = " << name_str << ":" << type << "[" << size << "]" << std::endl;
    }
}

void gl_context::enumerate_program_attributes(shader_program& prog, std::vector<std::string>& names, std::vector<int>* locations_ptr, std::vector<int>* sizes_ptr, std::vector<int>* types_ptr, bool show) const
{
    GLint count;
    glGetProgramiv(get_gl_id(prog.handle), GL_ACTIVE_ATTRIBUTES, &count);
    for (int i = 0; i < count; ++i) {
        GLchar name[1000];
        GLsizei length;
        GLint size;
        GLenum type;
        glGetActiveAttrib(get_gl_id(prog.handle), i, 1000, &length, &size, &type, name);
        std::string name_str(name, length);
        names.push_back(name_str);
        if (sizes_ptr)
            sizes_ptr->push_back(size);
        if (types_ptr)
            types_ptr->push_back(type);
        int loc = glGetAttribLocation(get_gl_id(prog.handle), name_str.c_str());
        if (locations_ptr)
            locations_ptr->push_back(loc);
        if (show)
            std::cout << i << " at " << loc << " = " << name_str << ":" << type << "[" << size << "]" << std::endl;
    }
}

void gl_context::set_color(const rgba& clr)
{
    current_color = clr;
    if (support_compatibility_mode && !core_profile) {
        glColor4fv(&clr[0]);
    }
    if (shader_program_stack.empty())
        return;
    cgv::render::shader_program& prog = *static_cast<cgv::render::shader_program*>(shader_program_stack.top());
    if (!prog.does_context_set_color())
        return;
    int clr_loc = prog.get_color_index();
    if (clr_loc == -1)
        return;
    prog.set_attribute(*this, clr_loc, clr);
}

void gl_context::set_material(const cgv::media::illum::surface_material& material)
{
    if (support_compatibility_mode && !core_profile) {
        unsigned side = map_to_gl(MS_FRONT_AND_BACK);
        float alpha = 1.0f - material.transparency;
        gl_set_material_color(side, material.ambient_occlusion * material.diffuse_reflectance, alpha, GL_AMBIENT);
        gl_set_material_color(side, material.diffuse_reflectance, alpha, GL_DIFFUSE);
        gl_set_material_color(side, material.specular_reflectance, alpha, GL_SPECULAR);
        gl_set_material_color(side, material.emission, alpha, GL_EMISSION);
        glMaterialf(side, GL_SHININESS, 1.0f / (material.roughness + 1.0f / 128.0f));
    }
    context::set_material(material);
}

void gl_context::enable_material(textured_material& mat)
{
    set_textured_material(mat);
    mat.enable_textures(*this); 
}

void gl_context::disable_material(textured_material& mat)
{
    mat.disable_textures(*this);
    current_material_ptr = 0;
    current_material_is_textured = false;
}
 
void gl_context::destruct_render_objects()
{
    for (unsigned i = 0; i < 4; ++i)
        progs[i].destruct(*this);
}

shader_program& gl_context::ref_default_shader_program(bool texture_support)
{
    if (!texture_support) {
        if (!progs[0].is_created()) {
            if (!progs[0].build_program(*this, "default.glpr")) {
                error("could not build default shader program from default.glpr");
                exit(0);
            }
            progs[0].specify_standard_uniforms(true, false, false, true);
            progs[0].specify_standard_vertex_attribute_names(*this, true, false, false);
            progs[0].allow_context_to_set_color(true);
        }
        return progs[0];
    }
    if (!progs[1].is_created()) {
        if (!progs[1].build_program(*this, "textured_default.glpr")) {
            error("could not build default shader program with texture support from textured_default.glpr");
            exit(0);
        }
        progs[1].set_uniform(*this, "texture", 0);
        progs[1].specify_standard_uniforms(true, false, false, true);
        progs[1].specify_standard_vertex_attribute_names(*this, true, false, true);
        progs[1].allow_context_to_set_color(true);
    }
    return progs[1];
}

shader_program& gl_context::ref_surface_shader_program(bool texture_support)
{
    if (!texture_support) {
        if (!progs[2].is_created()) {
            if (!progs[2].build_program(*this, "default_surface.glpr")) {
                error("could not build surface shader program from default_surface.glpr");
                exit(0);
            }
            progs[2].specify_standard_uniforms(true, true, true, true);
            progs[2].specify_standard_vertex_attribute_names(*this, true, true, false);
            progs[2].allow_context_to_set_color(true);
        }
        return progs[2];
    }
    if (!progs[3].is_created()) {
        if (!progs[3].build_program(*this, "textured_surface.glpr")) {
            error("could not build surface shader program with texture support from textured_surface.glpr");
            exit(0);
        }
        progs[3].specify_standard_uniforms(true, true, true, true);
        progs[3].specify_standard_vertex_attribute_names(*this, true, true, true);
        progs[3].allow_context_to_set_color(true);
    }
    return progs[3];
}
 
void gl_context::on_lights_changed()
{
    if (support_compatibility_mode && !core_profile) {
        GLint max_nr_lights;
        glGetIntegerv(GL_MAX_LIGHTS, &max_nr_lights);
        for (GLint light_idx = 0; light_idx < max_nr_lights; ++light_idx) {
            if (light_idx >= int(get_nr_enabled_light_sources())) {
                glDisable(GL_LIGHT0 + light_idx);
                continue;
            }
            GLfloat col[4] = { 1,1,1,1 };
            const cgv::media::illum::light_source& light = get_light_source(get_enabled_light_source_handle(light_idx));
            *(rgb*)col = light.get_emission()*light.get_ambient_scale();
            glLightfv(GL_LIGHT0 + light_idx, GL_AMBIENT, col);
            *(rgb*)col = light.get_emission();
            glLightfv(GL_LIGHT0 + light_idx, GL_DIFFUSE, col);
            *(rgb*)col = light.get_emission();
            glLightfv(GL_LIGHT0 + light_idx, GL_SPECULAR, col);
 
            GLfloat pos[4] = { 0,0,0,light.get_type() == cgv::media::illum::LT_DIRECTIONAL ? 0.0f : 1.0f };
            *(vec3*)pos = light.get_position();
            glLightfv(GL_LIGHT0 + light_idx, GL_POSITION, pos);
            if (light.get_type() != cgv::media::illum::LT_DIRECTIONAL) {
                glLightf(GL_LIGHT0 + light_idx, GL_CONSTANT_ATTENUATION, light.get_constant_attenuation());
                glLightf(GL_LIGHT0 + light_idx, GL_LINEAR_ATTENUATION, light.get_linear_attenuation());
                glLightf(GL_LIGHT0 + light_idx, GL_QUADRATIC_ATTENUATION, light.get_quadratic_attenuation());
            }
            else {
                glLightf(GL_LIGHT0 + light_idx, GL_CONSTANT_ATTENUATION, 1);
                glLightf(GL_LIGHT0 + light_idx, GL_LINEAR_ATTENUATION, 0);
                glLightf(GL_LIGHT0 + light_idx, GL_QUADRATIC_ATTENUATION, 0);
            }
            if (light.get_type() == cgv::media::illum::LT_SPOT) {
                glLightf(GL_LIGHT0 + light_idx, GL_SPOT_CUTOFF, light.get_spot_cutoff());
                glLightf(GL_LIGHT0 + light_idx, GL_SPOT_EXPONENT, light.get_spot_exponent());
                glLightfv(GL_LIGHT0 + light_idx, GL_SPOT_DIRECTION, light.get_spot_direction().data());
            }
            else {
                glLightf(GL_LIGHT0 + light_idx, GL_SPOT_CUTOFF, 180.0f);
                glLightf(GL_LIGHT0 + light_idx, GL_SPOT_EXPONENT, 0.0f);
                static float dir[3] = { 0,0,1 };
                glLightfv(GL_LIGHT0 + light_idx, GL_SPOT_DIRECTION, dir);
            }
            glEnable(GL_LIGHT0 + light_idx);
        }
    }
    context::on_lights_changed();
}
 
void gl_context::tesselate_arrow(double length, double aspect, double rel_tip_radius, double tip_aspect, int res, bool edges)
{
    double cyl_radius = length*aspect;
    double cone_radius = rel_tip_radius*cyl_radius;
    double cone_length = cone_radius/tip_aspect;
    double cyl_length = length - cone_length;
    push_modelview_matrix();
    mul_modelview_matrix(cgv::math::scale4(cyl_radius,cyl_radius,0.5*cyl_length));
        push_modelview_matrix();
            mul_modelview_matrix(cgv::math::rotate4(180.0,1.0,0.0,0.0));
            tesselate_unit_disk(res, false, edges);
        pop_modelview_matrix();
 
        mul_modelview_matrix(cgv::math::translate4(0.0,0.0,1.0));
        tesselate_unit_cylinder(res, false, edges);
 
    mul_modelview_matrix(cgv::math::translate4(0.0, 0.0, 1.0));
    mul_modelview_matrix(cgv::math::scale4(rel_tip_radius, rel_tip_radius, cone_length / cyl_length));
        push_modelview_matrix();
            mul_modelview_matrix(cgv::math::rotate4(180.0, 1.0, 0.0, 0.0));
                tesselate_unit_disk(res, false, edges);
        pop_modelview_matrix();
    mul_modelview_matrix(cgv::math::translate4(0.0, 0.0, 1.0));
        tesselate_unit_cone(res, false, edges);
    pop_modelview_matrix();
}

void gl_context::rotate_vector_to_target(const dvec3& vector, const dvec3& target)
{
    double angle;
    dvec3 axis;
    compute_rotation_axis_and_angle_from_vector_pair(vector, target, axis, angle);
    mul_modelview_matrix(cgv::math::rotate4<double>(180.0 / M_PI * angle, axis));
}

void gl_context::tesselate_arrow(const dvec3& start, const dvec3& end, double aspect, double rel_tip_radius, double tip_aspect, int res, bool edges)
{
    if ((start - end).length() < 1e-8) {
        error("ignored tesselate arrow called with start and end closer then 1e-8");
        return;
    }
    push_modelview_matrix();
        mul_modelview_matrix(cgv::math::translate4<double>(start));
        rotate_vector_to_target(dvec3(0, 0, 1), end - start);
        tesselate_arrow((end-start).length(), aspect, rel_tip_radius, tip_aspect, res, edges);
    pop_modelview_matrix();
}
 
void gl_context::draw_light_source(const light_source& l, float i, float light_scale)
{   
    set_color(i*l.get_emission());
    push_modelview_matrix();
    switch (l.get_type()) {
    case LT_DIRECTIONAL :
        mul_modelview_matrix(cgv::math::scale4<double>(light_scale, light_scale, light_scale));
        tesselate_arrow(vec3(0.0f), l.get_position(), 0.1f,2.0f,0.5f);
        break;
    case LT_POINT :
        mul_modelview_matrix(
            cgv::math::translate4<double>(l.get_position())*
            cgv::math::scale4<double>(vec3(0.3f*light_scale)));
        tesselate_unit_sphere();
        break;
    case LT_SPOT :
        mul_modelview_matrix(
            cgv::math::translate4<double>(l.get_position())*
            cgv::math::scale4<double>(vec3(light_scale))
        );
        rotate_vector_to_target(dvec3(0, 0, -1), l.get_spot_direction());
        {
            float t = tan(l.get_spot_cutoff()*(float)M_PI/180);
            if (l.get_spot_cutoff() > 45.0f)
                mul_modelview_matrix(cgv::math::scale4<double>(1, 1, 0.5f / t));
            else
                mul_modelview_matrix(cgv::math::scale4<double>(t, t, 0.5f));
            mul_modelview_matrix(cgv::math::translate4<double>(0, 0, -1));
            push_cull_state();
            set_cull_state(CM_OFF);
            tesselate_unit_cone();
            pop_cull_state();
        }
    }
    pop_modelview_matrix();
}
 
void gl_context::announce_external_frame_buffer_change(void*& fbo_handle)
{
    if (frame_buffer_stack.empty()) {
        error("gl_context::announce_external_frame_buffer_change() called with empty frame buffer stack");
        return;
    }
    GLint fbo_id = 0;
    glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &fbo_id);
    fbo_handle = frame_buffer_stack.top()->handle;
    frame_buffer_stack.top()->handle = get_handle(fbo_id);
}
 
void gl_context::recover_from_external_frame_buffer_change(void* fbo_handle)
{
    glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(fbo_handle));
    if (frame_buffer_stack.empty())
        return;
    frame_buffer_stack.top()->handle = fbo_handle;
}
 
void gl_context::announce_external_viewport_change(ivec4& cgv_viewport_storage)
{
    if (window_transformation_stack.empty()) {
        error("gl_context::announce_external_viewport_change() called with empty window transformation stack");
        return;
    }
    GLint vp[4];
    glGetIntegerv(GL_VIEWPORT, vp);
    cgv_viewport_storage = window_transformation_stack.top().front().viewport;
    window_transformation_stack.top().front().viewport = ivec4(vp[0], vp[1], vp[2], vp[3]);
}
 
void gl_context::recover_from_external_viewport_change(const ivec4& cgv_viewport_storage)
{
    glViewport(cgv_viewport_storage[0], cgv_viewport_storage[1], cgv_viewport_storage[2], cgv_viewport_storage[3]);
    if (window_transformation_stack.empty())
        return;
    window_transformation_stack.top().front().viewport = cgv_viewport_storage;
}

void gl_context::push_pixel_coords()
{
    push_projection_matrix();
    push_modelview_matrix();
 
    GLint vp[4];
    glGetIntegerv(GL_VIEWPORT, vp);
 
    if (support_compatibility_mode && !core_profile) {
        // push projection matrix
        glMatrixMode(GL_PROJECTION);
        // set orthogonal projection
        glLoadIdentity();
        glOrtho(vp[0], vp[0]+vp[2], vp[1], vp[1]+vp[3], -1, 1);
        // push modelview matrix
        glMatrixMode(GL_MODELVIEW);
        // use identity for modelview
        glLoadIdentity();
    }
    set_modelview_matrix(cgv::math::identity4<double>());
    set_projection_matrix(cgv::math::ortho4<double>(vp[0], vp[0] + vp[2], vp[1], vp[1]+vp[3], -1, 1));
}

void gl_context::pop_pixel_coords()
{
    pop_modelview_matrix();
    pop_projection_matrix();
}

bool gl_context::read_frame_buffer(data::data_view& dv, 
                                              unsigned int x, unsigned int y, FrameBufferType buffer_type, 
                                              TypeId type, data::ComponentFormat cf, int w, int h)
{
    const cgv::data::data_format* df = dv.get_format();
    if (df) {
        w = int(df->get_width());
        h = int(df->get_height());
        type = df->get_component_type();
        cf = df->get_standard_component_format();
        if (w < 1 || h < 1) {
            error(std::string("read_frame_buffer: received invalid dimensions (") + cgv::utils::to_string(w) + "," + cgv::utils::to_string(h) + ")");
            return false;
        }
    }
    else {
        if (w < 1 || h < 1) {
            GLint vp[4];
            glGetIntegerv(GL_VIEWPORT,vp);
            w = w < 1 ? vp[2] : w;
            h = h < 1 ? vp[3] : h;
        }
    }
    GLuint gl_type = map_to_gl(type);
    if (gl_type == 0) {
        error(std::string("read_frame_buffer: could not make component type ")+cgv::type::info::get_type_name(df->get_component_type())+" to gl type");
        return false;
    }
    GLuint gl_format = GL_DEPTH_COMPONENT;
    if (cf != cgv::data::CF_D) {
        gl_format = GL_STENCIL_INDEX;
        if (cf != cgv::data::CF_S) {
            gl_format = map_to_gl(cf);
            if (gl_format == GL_RGB && cf != cgv::data::CF_RGB) {
                error(std::string("read_frame_buffer: could not match component format ") + cgv::utils::to_string(df->get_component_format()));
                return false;
            }
        }
    }
    GLenum gl_buffer = GL_BACK;
    if (buffer_type < FB_BACK)
        gl_buffer = GL_COLOR_ATTACHMENT0+buffer_type;
    else {
        switch (buffer_type) {
        case FB_FRONT :       gl_buffer = GL_FRONT; break;
        case FB_BACK  :       gl_buffer = GL_BACK; break;
        case FB_FRONT_LEFT :  gl_buffer = GL_FRONT_LEFT; break;
        case FB_FRONT_RIGHT : gl_buffer = GL_FRONT_RIGHT; break;
        case FB_BACK_LEFT  :  gl_buffer = GL_BACK_LEFT; break;
        case FB_BACK_RIGHT  : gl_buffer = GL_BACK_RIGHT; break;
        default:
            error(std::string("invalid buffer type ")+cgv::utils::to_string(buffer_type));
            return false;
        }
    }
    // after all necessary information could be extracted, ensure that format 
    // and data view are created
    if (!df) {
        df = new cgv::data::data_format(w,h,type,cf);
        dv.~data_view();
        new(&dv) data::data_view(df);
        dv.manage_format(true);
    }
    GLint old_buffer, old_pack;
    glGetIntegerv(GL_READ_BUFFER, &old_buffer);
    glGetIntegerv(GL_PACK_ALIGNMENT, &old_pack);
    glReadBuffer(gl_buffer);
    glPixelStorei(GL_PACK_ALIGNMENT, df->get_component_alignment());
    glReadPixels(x,y,w,h,gl_format,gl_type,dv.get_ptr<void>());
    glReadBuffer(old_buffer);
    glPixelStorei(GL_PACK_ALIGNMENT, old_pack);
    dv.reflect_horizontally();
    return true;
}
 
 
void render_vertex(int k, const float* vertices, const float* normals, const float* tex_coords,
                          const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices, bool flip_normals)
{
    if (normals && normal_indices) {
        if (flip_normals) {
            float n[3] = { -normals[3*normal_indices[k]],-normals[3*normal_indices[k]+1],-normals[3*normal_indices[k]+2] };
            glNormal3fv(n);
        }
        else
            glNormal3fv(normals+3*normal_indices[k]);
    }
    if (tex_coords && tex_coord_indices)
        glTexCoord2fv(tex_coords+2*tex_coord_indices[k]);
    glVertex3fv(vertices+3*vertex_indices[k]);
}
 
attribute_array_binding*& get_aab_ptr()
{
    static attribute_array_binding* aab_ptr = 0;
    return aab_ptr;
}
 
vertex_buffer*& get_vbo_ptr()
{
    static vertex_buffer* vbo_ptr = 0;
    return vbo_ptr;
}
 
bool gl_context::release_attributes(const float* normals, const float* tex_coords, const int* normal_indices, const int* tex_coord_indices) const
{
    shader_program* prog_ptr = static_cast<shader_program*>(get_current_program());
    if (!prog_ptr || prog_ptr->get_position_index() == -1)
        return false;
 
    attribute_array_binding*& aab_ptr = get_aab_ptr();
    if (!aab_ptr) {
        attribute_array_binding::disable_global_array(*this, prog_ptr->get_position_index());
        if (prog_ptr->get_normal_index() != -1 && normals && normal_indices)
            attribute_array_binding::disable_global_array(*this, prog_ptr->get_normal_index());
        if (prog_ptr->get_texcoord_index() != -1 && tex_coords && tex_coord_indices)
            attribute_array_binding::disable_global_array(*this, prog_ptr->get_texcoord_index());
    }
    else {
        aab_ptr->disable(const_cast<gl_context&>(*this));
        aab_ptr->disable_array(*this, prog_ptr->get_position_index());
        if (prog_ptr->get_normal_index() != -1 && normals && normal_indices)
            aab_ptr->disable_global_array(*this, prog_ptr->get_normal_index());
        if (prog_ptr->get_texcoord_index() != -1 && tex_coords && tex_coord_indices)
            aab_ptr->disable_global_array(*this, prog_ptr->get_texcoord_index());
        vertex_buffer*& vbo_ptr = get_vbo_ptr();
        vbo_ptr->destruct(*this);
        delete vbo_ptr;
        vbo_ptr = 0;
    }
    return true;
}
 
bool gl_context::prepare_attributes(std::vector<vec3>& P, std::vector<vec3>& N, std::vector<vec2>& T, unsigned nr_vertices,
    const float* vertices, const float* normals, const float* tex_coords,
    const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices, bool flip_normals) const
{
    unsigned i;
    shader_program* prog_ptr = static_cast<shader_program*>(get_current_program());
    if (!prog_ptr || prog_ptr->get_position_index() == -1)
        return false;
    P.resize(nr_vertices);
    for (i = 0; i < nr_vertices; ++i)
        P[i] = *reinterpret_cast<const vec3*>(vertices + 3 * vertex_indices[i]);
 
    if (prog_ptr->get_normal_index() != -1 && normals && normal_indices) {
        N.resize(nr_vertices);
        for (i = 0; i < nr_vertices; ++i) {
            N[i] = *reinterpret_cast<const vec3*>(normals + 3 * normal_indices[i]);
            if (flip_normals)
                N[i] = -N[i];
        }
    }
    if (prog_ptr->get_texcoord_index() != -1 && tex_coords && tex_coord_indices) {
        T.resize(nr_vertices);
        for (i = 0; i < nr_vertices; ++i)
            T[i] = *reinterpret_cast<const vec2*>(tex_coords + 2 * tex_coord_indices[i]);
    }
 
    attribute_array_binding*& aab_ptr = get_aab_ptr();
    if (core_profile && !aab_ptr) {
        aab_ptr = new attribute_array_binding();
        aab_ptr->create(*this);
    }
    if (!aab_ptr) {
        attribute_array_binding::set_global_attribute_array<>(*this, prog_ptr->get_position_index(), P);
        attribute_array_binding::enable_global_array(*this, prog_ptr->get_position_index());
        if (prog_ptr->get_normal_index() != -1) {
            if (normals && normal_indices) {
                attribute_array_binding::set_global_attribute_array<>(*this, prog_ptr->get_normal_index(), N);
                attribute_array_binding::enable_global_array(*this, prog_ptr->get_normal_index());
            }
            else
                attribute_array_binding::disable_global_array(*this, prog_ptr->get_normal_index());
        }
        if (prog_ptr->get_texcoord_index() != -1) {
            if (tex_coords && tex_coord_indices) {
                attribute_array_binding::set_global_attribute_array<>(*this, prog_ptr->get_texcoord_index(), T);
                attribute_array_binding::enable_global_array(*this, prog_ptr->get_texcoord_index());
            }
            else
                attribute_array_binding::disable_global_array(*this, prog_ptr->get_texcoord_index());
        }
    }
    else {
        vertex_buffer*& vbo_ptr = get_vbo_ptr();
        if (!vbo_ptr)
            vbo_ptr = new vertex_buffer();
        else
            vbo_ptr->destruct(*this);
        vbo_ptr->create(*this, P.size() * sizeof(vec3) + N.size() * sizeof(vec3) + T.size() * sizeof(vec2));
        vbo_ptr->replace(const_cast<gl_context&>(*this), 0, P.data(), P.size());
        size_t nml_off = P.size() * sizeof(vec3);
        size_t tex_off = nml_off;
        if (!N.empty()) {
            vbo_ptr->replace(const_cast<gl_context&>(*this), nml_off, N.data(), N.size());
            tex_off += N.size() * sizeof(vec3);
        }
        if (!T.empty())
            vbo_ptr->replace(const_cast<gl_context&>(*this), tex_off, T.data(), T.size());
 
        type_descriptor td3 = element_descriptor_traits<vec3>::get_type_descriptor(P.front());
        aab_ptr->set_attribute_array(*this, prog_ptr->get_position_index(), td3, *vbo_ptr, 0, P.size());
        aab_ptr->enable_array(*this, prog_ptr->get_position_index());
        if (prog_ptr->get_normal_index() != -1) {
            if (normals && normal_indices) {
                aab_ptr->set_attribute_array(*this, prog_ptr->get_normal_index(), td3, *vbo_ptr, nml_off, N.size());
                aab_ptr->enable_array(*this, prog_ptr->get_normal_index());
            }
            else
                aab_ptr->disable_array(*this, prog_ptr->get_normal_index());
        }
        if (prog_ptr->get_texcoord_index() != -1) {
            if (tex_coords && tex_coord_indices) {
                type_descriptor td2 = element_descriptor_traits<vec2>::get_type_descriptor(T.front());
                aab_ptr->set_attribute_array(*this, prog_ptr->get_texcoord_index(), td2, *vbo_ptr, tex_off, T.size());
                aab_ptr->enable_array(*this, prog_ptr->get_texcoord_index());
            }
            else
                aab_ptr->disable_array(*this, prog_ptr->get_texcoord_index());
        }
        aab_ptr->enable(const_cast<gl_context&>(*this));
    }
    return true;
}

void gl_context::draw_edges_of_faces(
    const float* vertices, const float* normals, const float* tex_coords,
    const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices,
    int nr_faces, int face_degree, bool flip_normals) const
{
    if (draw_in_compatibility_mode) {
        int k = 0;
        for (int i = 0; i < nr_faces; ++i) {
            glBegin(GL_LINE_LOOP);
            for (int j = 0; j < face_degree; ++j, ++k)
                render_vertex(k, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
            glEnd();
        }
        return;
    }
    unsigned nr_vertices = face_degree * nr_faces;
    std::vector<vec3> P, N;
    std::vector<vec2> T;
    if (!prepare_attributes(P, N, T, nr_vertices, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals))
        return;
    for (int i = 0; i < nr_faces; ++i)
        glDrawArrays(GL_LINE_LOOP, i*face_degree, face_degree);
    release_attributes(normals, tex_coords, normal_indices, tex_coord_indices);
}
 
 
void gl_context::draw_elements_void(GLenum mode, size_t total_count, GLenum type, size_t type_size, const void* indices) const
{
    ensure_configured();
    size_t drawn = 0;
    const cgv::type::uint8_type* index_ptr = static_cast<const cgv::type::uint8_type*>(indices);
    while (drawn < total_count) {
        size_t count = total_count - drawn;
        if (count > max_nr_indices)
            count = size_t(max_nr_indices);
        glDrawElements(mode, GLsizei(count), type, index_ptr + drawn * type_size);
        drawn += count;
    }
}
 
size_t max_nr_indices, max_nr_vertices;
void gl_context::ensure_configured() const
{
    if (max_nr_indices != 0)
        return;
    glGetInteger64v(GL_MAX_ELEMENTS_INDICES, reinterpret_cast<GLint64*>(&max_nr_indices));
    glGetInteger64v(GL_MAX_ELEMENTS_VERTICES, reinterpret_cast<GLint64*>(&max_nr_vertices));
}

void gl_context::draw_edges_of_strip_or_fan(
    const float* vertices, const float* normals, const float* tex_coords,
    const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices,
    int nr_faces, int face_degree, bool is_fan, bool flip_normals) const
{
    int s = face_degree - 2;
    int i, k = 2;
    std::vector<GLuint> I;
    I.push_back(0); I.push_back(1);
    for (i = 0; i < nr_faces; ++i) {
        if (is_fan) {
            I.push_back(k - 1); I.push_back(k);
            I.push_back(0); I.push_back(k);
            continue;
        }
        if (s == 1) {
            I.push_back(k - 1); I.push_back(k);
            I.push_back(k - 2); I.push_back(k);
        }
        else {
            I.push_back(k - 1); I.push_back(k + 1);
            I.push_back(k - 2); I.push_back(k);
            I.push_back(k); I.push_back(k + 1);
        }
        k += 2;
    }
 
    if (draw_in_compatibility_mode) {
        glBegin(GL_LINES);
        for (GLuint j:I)
            render_vertex(j, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
        glEnd();
        return;
    }
    unsigned nr_vertices = 2 + (face_degree - 2) * nr_faces;
    std::vector<vec3> P, N;
    std::vector<vec2> T;
    if (!prepare_attributes(P, N, T, nr_vertices, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals))
        return;
    draw_elements(GL_LINES, I.size(), &I[0]);
    release_attributes(normals, tex_coords, normal_indices, tex_coord_indices);
}
 
void gl_context::draw_faces(
    const float* vertices, const float* normals, const float* tex_coords, 
    const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices, 
    int nr_faces, int face_degree, bool flip_normals) const
{
    if (draw_in_compatibility_mode) {
        int k = 0;
        if (face_degree < 5)
            glBegin(face_degree == 3 ? GL_TRIANGLES : GL_QUADS);
        for (int i = 0; i < nr_faces; ++i) {
            if (face_degree >= 5)
                glBegin(GL_POLYGON);
            for (int j = 0; j < face_degree; ++j, ++k)
                render_vertex(k, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
            if (face_degree >= 5)
                glEnd();
        }
        if (face_degree < 5)
            glEnd();
        return;
    }
    unsigned nr_vertices = face_degree * nr_faces;
    std::vector<vec3> P, N;
    std::vector<vec2> T;
    if (!prepare_attributes(P, N, T, nr_vertices, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals))
        return;
    /*
    for (unsigned x = 0; x < nr_vertices; ++x) {
        std::cout << x << ": [" << P[x] << "]";
        if (N.size() > 0)
            std::cout << ", <" << N[x] << ">";
        if (T.size() > 0) 
            std::cout << ", {" << T[x] << "}";
        std::cout << std::endl;
    }
    */
    if (face_degree == 3)
        glDrawArrays(GL_TRIANGLES, 0, nr_vertices);
    else {
        for (int i = 0; i < nr_faces; ++i) {
            glDrawArrays(GL_TRIANGLE_FAN, i*face_degree, face_degree);
        }
    }
    release_attributes(normals, tex_coords, normal_indices, tex_coord_indices);
}
 
void gl_context::draw_strip_or_fan(
        const float* vertices, const float* normals, const float* tex_coords, 
        const int* vertex_indices, const int* normal_indices, const int* tex_coord_indices, 
        int nr_faces, int face_degree, bool is_fan, bool flip_normals) const
{
    int s = face_degree - 2;
    int k = 2;
    if (draw_in_compatibility_mode) {
        glBegin(face_degree == 3 ? (is_fan ? GL_TRIANGLE_FAN : GL_TRIANGLE_STRIP) : GL_QUAD_STRIP);
        render_vertex(0, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
        render_vertex(1, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
        for (int i = 0; i < nr_faces; ++i)
            for (int j = 0; j < s; ++j, ++k)
                render_vertex(k, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals);
        glEnd();
        return;
    }
    unsigned nr_vertices = 2 + (face_degree-2) * nr_faces;
    std::vector<vec3> P, N;
    std::vector<vec2> T;
    if (!prepare_attributes(P, N, T, nr_vertices, vertices, normals, tex_coords, vertex_indices, normal_indices, tex_coord_indices, flip_normals))
        return;
    glDrawArrays(is_fan ? GL_TRIANGLE_FAN : GL_TRIANGLE_STRIP, 0, nr_vertices);
    release_attributes(normals, tex_coords, normal_indices, tex_coord_indices);
}
 
void gl_context::set_depth_test_state(DepthTestState state) {
    if(state.enabled)
        glEnable(GL_DEPTH_TEST);
    else
        glDisable(GL_DEPTH_TEST);
    glDepthFunc(map_to_gl(state.test_func));
    context::set_depth_test_state(state);
}
 
void gl_context::set_depth_func(CompareFunction func) {
    glDepthFunc(map_to_gl(func));
    context::set_depth_func(func);
}
 
void gl_context::enable_depth_test() {
    glEnable(GL_DEPTH_TEST);
    context::enable_depth_test();
}
 
void gl_context::disable_depth_test() {
    glDisable(GL_DEPTH_TEST);
    context::disable_depth_test();
}
 
void gl_context::set_cull_state(CullingMode culling_mode) {
    switch(culling_mode) {
    case CM_OFF:
        glDisable(GL_CULL_FACE);
        break;
    case CM_BACKFACE:
        glEnable(GL_CULL_FACE);
        glCullFace(GL_BACK);
        break;
    case CM_FRONTFACE:
        glEnable(GL_CULL_FACE);
        glCullFace(GL_FRONT);
        break;
    default:
        break;
    }
    context::set_cull_state(culling_mode);
}
 
void gl_context::set_blend_state(BlendState state) {
    if(state.enabled)
        glEnable(GL_BLEND);
    else
        glDisable(GL_BLEND);
    if(GLEW_EXT_blend_func_separate) {
        glBlendFuncSeparate(
            map_to_gl(state.src_color),
            map_to_gl(state.dst_color),
            map_to_gl(state.src_alpha),
            map_to_gl(state.dst_alpha)
        );
    } else {
        glBlendFunc(map_to_gl(state.src_color), map_to_gl(state.dst_color));
    }
    context::set_blend_state(state);
}
 
void gl_context::set_blend_func(BlendFunction src_factor, BlendFunction dst_factor) {
    glBlendFunc(map_to_gl(src_factor), map_to_gl(dst_factor));
    context::set_blend_func(src_factor, dst_factor);
}
 
void gl_context::set_blend_func_separate(BlendFunction src_color_factor, BlendFunction dst_color_factor, BlendFunction src_alpha_factor, BlendFunction dst_alpha_factor) {
    glBlendFuncSeparate(
        map_to_gl(src_color_factor),
        map_to_gl(dst_color_factor),
        map_to_gl(src_alpha_factor),
        map_to_gl(dst_alpha_factor)
    );
    context::set_blend_func_separate(src_color_factor, dst_color_factor, src_alpha_factor, dst_alpha_factor);
}
 
void gl_context::enable_blending() {
    glEnable(GL_BLEND);
    context::enable_blending();
}
 
void gl_context::disable_blending() {
    glDisable(GL_BLEND);
    context::disable_blending();
}
 
void gl_context::set_buffer_mask(BufferMask mask) {
    glDepthMask(map_to_gl(mask.depth_flag));
    glColorMask(
        map_to_gl(mask.red_flag),
        map_to_gl(mask.green_flag),
        map_to_gl(mask.blue_flag),
        map_to_gl(mask.alpha_flag)
    );
    context::set_buffer_mask(mask);
}
 
void gl_context::set_depth_mask(bool flag) {
    glDepthMask(map_to_gl(flag));
    context::set_depth_mask(flag);
}
 
void gl_context::set_color_mask(bvec4 flags) {
    glColorMask(
        map_to_gl(flags[0]),
        map_to_gl(flags[1]),
        map_to_gl(flags[2]),
        map_to_gl(flags[3])
    );
    context::set_color_mask(flags);
}

dmat4 gl_context::get_modelview_matrix() const
{
    if (support_compatibility_mode && !core_profile) {
        GLdouble V[16];
        glGetDoublev(GL_MODELVIEW_MATRIX, V);
        return dmat4(4,4,V);
    }
    if (modelview_matrix_stack.empty())
        return identity4<double>();
    return modelview_matrix_stack.top();
}

dmat4 gl_context::get_projection_matrix() const
{
    if (support_compatibility_mode && !core_profile) {
        GLdouble P[16];
        glGetDoublev(GL_PROJECTION_MATRIX, P);
        return dmat4(4,4,P);
    }
    if (projection_matrix_stack.empty())
        return identity4<double>();
    return projection_matrix_stack.top();
}

void gl_context::pop_window_transformation_array()
{
    context::pop_window_transformation_array();
    update_window_transformation_array();
}
 
void gl_context::update_window_transformation_array()
{
    const std::vector<window_transformation>& wta = window_transformation_stack.top();
    if (wta.size() == 1) {
        const ivec4& viewport = wta.front().viewport;
        const dvec2& depth_range = wta.front().depth_range;
        glViewport(viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
        glScissor(viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
        glDepthRange(depth_range[0], depth_range[1]);
    }
    else {
        for (GLuint array_index = 0; array_index < (GLuint)wta.size(); ++array_index) {
            const ivec4& viewport    = wta[array_index].viewport;
            const dvec2& depth_range = wta[array_index].depth_range;
            glViewportIndexedf(array_index, (GLfloat)viewport[0], (GLfloat)viewport[1], (GLfloat)viewport[2], (GLfloat)viewport[3]);
            glScissorIndexed(array_index, viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
            glDepthRangeIndexed(array_index, (GLclampd)depth_range[0], (GLclampd)depth_range[1]);
        }
    }
}
 
unsigned gl_context::get_max_window_transformation_array_size() const
{
    GLint max_value = 1;
    if (GLEW_VERSION_4_1)
        glGetIntegerv(GL_MAX_VIEWPORTS, &max_value);
    return max_value;
}

void gl_context::set_viewport(const ivec4& viewport, int array_index)
{
    size_t nr = window_transformation_stack.top().size();
    context::set_viewport(viewport, array_index);
    if (nr != window_transformation_stack.top().size())
        update_window_transformation_array();
    else {
        if (array_index < 1) {
            glViewport(viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
            glScissor(viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
        }
        else {
            glViewportIndexedf(array_index, (GLfloat)viewport[0], (GLfloat)viewport[1], (GLfloat)viewport[2], (GLfloat)viewport[3]);
            glScissorIndexed(array_index, viewport[0], viewport[1], (GLsizei)viewport[2], (GLsizei)viewport[3]);
        }
    }
}

void gl_context::set_depth_range(const dvec2& depth_range, int array_index)
{
    size_t nr = window_transformation_stack.top().size();
    context::set_depth_range(depth_range, array_index);
    if (nr != window_transformation_stack.top().size())
        update_window_transformation_array();
    else {
        if (array_index < 1)
            glDepthRange(depth_range[0], depth_range[1]);
        else
            glDepthRangeIndexed(array_index, (GLclampd)depth_range[0], (GLclampd)depth_range[1]);
    }
}
/*
gl_context::dmat4 gl_context::get_device_matrix() const
{
    GLint vp[4];
    glGetIntegerv(GL_VIEWPORT, vp);
    dmat4 D; D.zeros();
    D(0, 0) = 0.5*vp[2];
    D(0, 3) = 0.5*vp[2] + vp[0];
    D(1, 1) = -0.5*vp[3]; // flip y-coordinate
    D(1, 3) = get_height() - 0.5*vp[3] - vp[1];
    D(2, 2) = 0.5;
    D(2, 3) = 0.5;
    D(3, 3) = 1.0;
    return D;
}
*/
void gl_context::set_modelview_matrix(const dmat4& V)
{
    if (support_compatibility_mode && !core_profile) {
        GLint mm;
        glGetIntegerv(GL_MATRIX_MODE, &mm);
        glMatrixMode(GL_MODELVIEW);
        glLoadMatrixd(V.data());
        glMatrixMode(mm);
    }
    context::set_modelview_matrix(V);
}
 
void gl_context::set_projection_matrix(const dmat4& P)
{
    if (support_compatibility_mode && !core_profile) {
        GLint mm;
        glGetIntegerv(GL_MATRIX_MODE, &mm);
        glMatrixMode(GL_PROJECTION);
        glLoadMatrixd(P.data());
        glMatrixMode(mm);
    }
    context::set_projection_matrix(P);
}

double gl_context::get_window_z(int x_window, int y_window) const
{
    GLfloat z_window;
 
    if (!in_render_process() && !is_current())
        make_current();
 
    glReadPixels(x_window, y_window, 1, 1, GL_DEPTH_COMPONENT, GL_FLOAT, &z_window);
    /*
    GLenum err = glGetError();
    if (err != GL_NO_ERROR) {
        std::cout << "gl error:";
        switch (err) {
            case GL_INVALID_ENUM :      std::cout << "invalid enum"; break;
            case GL_INVALID_VALUE :     std::cout << "invalid value"; break;
            case GL_INVALID_OPERATION : std::cout << "invalid operation"; break;
            case GL_STACK_OVERFLOW :    std::cout << "stack overflow"; break;
            case GL_STACK_UNDERFLOW :   std::cout << "stack underflow"; break;
            case GL_OUT_OF_MEMORY :     std::cout << "out of memory"; break;
            default:                    std::cout << "unknown error"; break;
        }
        std::cout << std::endl;
 
    }
    */
    return z_window;
}
 
#include <iomanip>
cgv::data::component_format gl_context::texture_find_best_format(
                const cgv::data::component_format& cf, 
                render_component& rc, const std::vector<cgv::data::data_view>* palettes) const
{
    GLuint& gl_format = (GLuint&)rc.internal_format;
    cgv::data::component_format best_cf;
    if (debug_texture_format_matching)
        std::cout << "Texture Format Search for '" << cf << "':" << std::endl;
    gl_format = find_best_texture_format(cf, &best_cf, palettes, debug_texture_format_matching);
    if (debug_texture_format_matching)
        std::cout << "found: '" << best_cf << "' = " << std::hex << gl_format << std::endl;
    return best_cf;
}
 
std::string gl_error_to_string(GLenum eid) {
    switch (eid) {
    case GL_NO_ERROR: return "";
    case GL_INVALID_ENUM: return "invalid enum";
    case GL_INVALID_VALUE: return "invalid value";
    case GL_INVALID_OPERATION: return "invalid operation";
    case GL_INVALID_FRAMEBUFFER_OPERATION: return "invalid framebuffe";
    case GL_OUT_OF_MEMORY: return "out of memory";
    case GL_STACK_UNDERFLOW: return "stack underflow";
    case GL_STACK_OVERFLOW: return "stack overflow";
    default: 
        return "undefined error (id: " + std::to_string(eid) + ")";
    }
    //return std::string((const char*)gluErrorString(eid));
}
 
std::string gl_error() {
    GLenum eid = glGetError();
    // Clear potential other error flags, see https://registry.khronos.org/OpenGL-Refpages/gl4/html/glGetError.xhtml:
    // "glGetError should always be called in a loop, until it returns GL_NO_ERROR, if all error flags are to be reset."
    while (glGetError());
    return gl_error_to_string(eid);
}
 
bool gl_context::check_gl_error(const std::string& where, const cgv::render::render_component* rc) const
{
    GLenum eid = glGetError();
    if (eid == GL_NO_ERROR)
        return false;
    std::string error_string = where + ": " + gl_error_to_string(eid);
    error(error_string, rc);
    while (glGetError()); // Clear all error flags, see gl_error.
    return true;
}
 
bool gl_context::check_texture_support(TextureType tt, const std::string& where, const cgv::render::render_component* rc) const
{
    switch (tt) {
    case TT_3D:
        if (!GLEW_VERSION_1_2) {
            error(where + ": 3D texture not supported", rc);
            return false;
        }
        break;
    case TT_CUBEMAP:
        if (!GLEW_VERSION_1_3) {
            error(where + ": cubemap texture not supported", rc);
            return false;
        }
    default:
        break;
    }
    return true;
}
 
bool gl_context::check_shader_support(ShaderType st, const std::string& where, const cgv::render::render_component* rc) const
{
    switch (st) {
    case ST_TASK:
    case ST_MESH:
        if(GLEW_VERSION_4_5) {
            if(GLEW_NV_mesh_shader || GLEW_EXT_mesh_shader)
                return true;
            else {
                error(where + ": task and mesh shaders require support of the GL_NV_mesh_shader or GLSL_EXT_mesh_shader extensions", rc);
                return false;
            }
        } else {
            error(where + ": task and mesh shaders are only supported starting from OpenGL version 4.5", rc);
            return false;
        }
    case ST_COMPUTE:
        if (GLEW_VERSION_4_3)
            return true;
        else {
            error(where+": compute shaders are only supported starting from OpenGL version 4.3", rc);
            return false;
        }
    case ST_TESS_CONTROL:
    case ST_TESS_EVALUATION:
        if (GLEW_VERSION_4_0)
            return true;
        else {
            error(where+": tessellation shaders are only supported starting from OpenGL version 4.0", rc);
            return false;
        }
    case ST_GEOMETRY:
        if (GLEW_VERSION_3_2)
            return true;
        else {
            error(where + ": geometry shaders are only supported starting from OpenGL version 3.2", rc);
            return false;
        }
    default:
        if (GLEW_VERSION_2_0)
            return true;
        else {
            error(where + ": shaders are only supported starting from OpenGL version 2.0", rc);
            return false;
        }
    }
}
 
bool gl_context::check_fbo_support(const std::string& where, const cgv::render::render_component* rc) const
{
    if (!GLEW_VERSION_3_0) {
        error(where + ": framebuffer objects not supported", rc);
        return false;
    }
    return true;
}
 
GLuint gl_context::texture_generate(texture_base& tb) const
{
    if (!check_texture_support(tb.tt, "gl_context::texture_generate", &tb))
        return get_gl_id(0);
    GLuint tex_id = get_gl_id(0);
    glGenTextures(1, &tex_id);
    if (glGetError() == GL_INVALID_OPERATION)
        error("gl_context::texture_generate: attempt to create texture inside glBegin-glEnd-block", &tb);
    return tex_id;
}
 
GLuint gl_context::texture_bind(TextureType tt, GLuint tex_id) const
{
    GLint tmp_id;
    glGetIntegerv(get_tex_bind(tt), &tmp_id);
    glBindTexture(get_tex_dim(tt), tex_id);
    return tmp_id;
}
 
void gl_context::texture_unbind(TextureType tt, GLuint tmp_id) const
{
    glBindTexture(get_tex_dim(tt), tmp_id);
}
 
bool gl_context::texture_create(texture_base& tb, cgv::data::data_format& df) const
{
    GLuint gl_format = (const GLuint&) tb.internal_format;
    
    if (tb.tt == TT_UNDEF)
        tb.tt = (TextureType)df.get_nr_dimensions();
    GLuint tex_id = texture_generate(tb);
    if (tex_id == get_gl_id(0))
        return false;
    GLuint tmp_id = texture_bind(tb.tt, tex_id);
 
    // extract component type
    unsigned int transfer_format = map_to_gl(df.get_standard_component_format(), df.get_integer_interpretation());
    if (transfer_format == -1) {
        error("could not determine transfer format", &tb);
        return false;
    }
    switch (tb.tt) {
    case TT_1D :
        glTexImage1D(GL_TEXTURE_1D, 0, 
            gl_format, GLsizei(df.get_width()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        break;
    case TT_1D_ARRAY :
        glTexImage2D(GL_TEXTURE_1D_ARRAY, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        break;
    case TT_2D :
        glTexImage2D(GL_TEXTURE_2D, 0, gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        break;
    case TT_MULTISAMPLE_2D:
        glTexImage2DMultisample(GL_TEXTURE_2D_MULTISAMPLE, tb.nr_multi_samples, gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), map_to_gl(tb.fixed_sample_locations));
        break;
    case TT_2D_ARRAY :
        glTexImage3D(GL_TEXTURE_2D_ARRAY, 0, gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), GLsizei(df.get_depth()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        break;
    case TT_MULTISAMPLE_2D_ARRAY:
        glTexStorage3DMultisample(GL_TEXTURE_2D_MULTISAMPLE_ARRAY, tb.nr_multi_samples, gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), GLsizei(df.get_depth()), map_to_gl(tb.fixed_sample_locations));
        break;
    case TT_3D :
        glTexImage3D(GL_TEXTURE_3D, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), GLsizei(df.get_depth()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        break;
    case TT_CUBEMAP :
        glTexImage2D( GL_TEXTURE_CUBE_MAP_POSITIVE_X, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D( GL_TEXTURE_CUBE_MAP_NEGATIVE_X, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D( GL_TEXTURE_CUBE_MAP_POSITIVE_Y, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D( GL_TEXTURE_CUBE_MAP_NEGATIVE_Y, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D( GL_TEXTURE_CUBE_MAP_POSITIVE_Z, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D( GL_TEXTURE_CUBE_MAP_NEGATIVE_Z, 0,
            gl_format, GLsizei(df.get_width()), GLsizei(df.get_height()), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
    default:
        break;
    }
    if (check_gl_error("gl_context::texture_create", &tb)) {
        glDeleteTextures(1, &tex_id);
        texture_unbind(tb.tt, tmp_id);
        return false;
    }
 
    texture_unbind(tb.tt, tmp_id);
    tb.have_mipmaps = false;
    tb.handle = get_handle(tex_id);
    return true;
}
 
bool gl_context::texture_create(
                            texture_base& tb,
                            cgv::data::data_format& target_format,
                            const cgv::data::const_data_view& data,
                            int level, int cube_side, int num_array_layers, const std::vector<cgv::data::data_view>* palettes) const
{
    // query the format to be used for the texture
    GLuint gl_tex_format = (const GLuint&) tb.internal_format;
 
    // define the texture type from the data format and the cube_side parameter
    tb.tt = (TextureType)data.get_format()->get_nr_dimensions();
    if(cube_side > -1) {
        if(tb.tt == TT_2D)
            tb.tt = TT_CUBEMAP;
    } else if(num_array_layers != 0) {
        if(num_array_layers < 0) {
            // automatic inference of layers from texture dimensions
            unsigned n_dims = data.get_format()->get_nr_dimensions();
            if(n_dims == 2)
                tb.tt = TT_1D_ARRAY;
            if(n_dims == 3)
                tb.tt = TT_2D_ARRAY;
        } else {
            switch(tb.tt) {
            case TT_1D: tb.tt = TT_1D_ARRAY; break;
            case TT_2D: tb.tt = TT_2D_ARRAY; break;
            case TT_3D: tb.tt = TT_2D_ARRAY; break;
            default: 
                break;
            }
        }
    }
    // create texture is not yet done
    GLuint tex_id;
    if (tb.is_created()) 
        tex_id = get_gl_id(tb.handle);
    else {
        tex_id = texture_generate(tb);
        if (tex_id == get_gl_id(0))
            return false;
        tb.handle = get_handle(tex_id);
    }
 
    // bind texture
    GLuint tmp_id = texture_bind(tb.tt, tex_id);
 
    // load data to texture
    tb.have_mipmaps = load_texture(data, gl_tex_format, level, cube_side, num_array_layers, palettes);
    bool result = !check_gl_error("gl_context::texture_create", &tb);
    // restore old texture
    texture_unbind(tb.tt, tmp_id);
    return result;
}
 
bool gl_context::texture_create_from_buffer(
                        texture_base& tb, 
                        cgv::data::data_format& df, 
                        int x, int y, int level) const
{
    GLuint gl_format = (const GLuint&) tb.internal_format;
 
    tb.tt = (TextureType)df.get_nr_dimensions();
    if (tb.tt != TT_2D) {
        tb.last_error = "texture creation from buffer only possible for 2d textures";
        return false;
    }
    GLuint tex_id;
    if (tb.is_created()) 
        tex_id = get_gl_id(tb.handle);
    else {
        tex_id = texture_generate(tb);
        if (tex_id == get_gl_id(0))
            return false;
        tb.handle = get_handle(tex_id);
    }
    GLuint tmp_id = texture_bind(tb.tt, tex_id);
 
    // check mipmap type
    bool gen_mipmap = level == -1;
    if (gen_mipmap)
        level = 0;
 
    glCopyTexImage2D(GL_TEXTURE_2D, level, gl_format, x, y, GLsizei(df.get_width()), GLsizei(df.get_height()), 0);
    bool result = false;
    std::string error_string("gl_context::texture_create_from_buffer: ");
    switch (glGetError()) {
    case GL_NO_ERROR :
        result = true;
        break;
    case GL_INVALID_ENUM : 
        error_string += "target was not an accepted value."; 
        break;
    case GL_INVALID_VALUE : 
        error_string += "level was less than zero or greater than log sub 2(max), where max is the returned value of GL_MAX_TEXTURE_SIZE.\n"
                             "or border was not zero or 1.\n"
                             "or width was less than zero, greater than 2 + GL_MAX_TEXTURE_SIZE; or width cannot be represented as 2n + 2 * border for some integer n.";
        break;
    case GL_INVALID_OPERATION : 
        error_string += "glCopyTexImage2D was called between a call to glBegin and the corresponding call to glEnd.";
        break;
    default:
        error_string += gl_error_to_string(glGetError());
        break;
    }
    texture_unbind(tb.tt, tmp_id);
    if (!result)
        error(error_string, &tb);
    else
        if (gen_mipmap)
            result = texture_generate_mipmaps(tb, tb.tt == TT_CUBEMAP ? 2 : (int)tb.tt);
 
    return result;
}
 
bool gl_context::texture_replace(
                        texture_base& tb, 
                        int x, int y, int z, 
                        const cgv::data::const_data_view& data, 
                        int level, const std::vector<cgv::data::data_view>* palettes) const
{
    if (!tb.is_created()) {
        error("gl_context::texture_replace: attempt to replace in not created texture", &tb);
        return false;
    }
    // determine dimension from location arguments
    unsigned int dim = 1;
    if (y != -1) {
        ++dim;
        if (z != -1)
            ++dim;
    }
    // check consistency
    if (tb.tt == TT_CUBEMAP) {
        if (dim != 3) {
            error("gl_context::texture_replace: replace on cubemap without the side defined", &tb);
            return false;
        }
        if (z < 0 || z > 5) {
            error("gl_context::texture_replace: replace on cubemap with invalid side specification", &tb);
            return false;
        }
    }
    else {
        if (tb.tt != dim) {
            error("gl_context::texture_replace: replace on texture with invalid position specification", &tb);
            return false;
        }
    }
 
    // bind texture
    GLuint tmp_id = texture_bind(tb.tt, get_gl_id(tb.handle));
    tb.have_mipmaps = replace_texture(data, level, x, y, z, palettes) || tb.have_mipmaps;
    bool result = !check_gl_error("gl_context::texture_replace", &tb);
    texture_unbind(tb.tt, tmp_id);
    return result;
}
 
bool gl_context::texture_replace_from_buffer(
                            texture_base& tb, 
                            int x, int y, int z, 
                            int x_buffer, int y_buffer, 
                            unsigned int width, unsigned int height, 
                            int level) const
{
    if (!tb.is_created()) {
        error("gl_context::texture_replace_from_buffer: attempt to replace in not created texture", &tb);
        return false;
    }
    // determine dimension from location arguments
    unsigned int dim = 2;
    if (z != -1)
        ++dim;
 
    // consistency checks
    if (tb.tt == TT_CUBEMAP) {
        if (dim != 3) {
            error("gl_context::texture_replace_from_buffer: replace on cubemap without the side defined", &tb);
            return false;
        }
        if (z < 0 || z > 5) {
            error("gl_context::texture_replace_from_buffer: replace on cubemap without invalid side specification", &tb);
            return false;
        }
    }
    else {
        if (tb.tt != dim) {
            tb.last_error = "replace on texture with invalid position specification";
            return false;
        }
    }
    // check mipmap type
    bool gen_mipmap = level == -1;
    if (gen_mipmap)
        level = 0;
 
    // bind texture
    GLuint tmp_id = texture_bind(tb.tt, get_gl_id(tb.handle));
    switch (tb.tt) {
    case TT_2D :      glCopyTexSubImage2D(GL_TEXTURE_2D, level, x, y, x_buffer, y_buffer, width, height); break;
    case TT_3D :      glCopyTexSubImage3D(GL_TEXTURE_3D, level, x, y, z, x_buffer, y_buffer, width, height); break;
    case TT_CUBEMAP : glCopyTexSubImage2D(get_gl_cube_map_target(z), level, x, y, x_buffer, y_buffer, width, height); 
    default: break;
    }
    bool result = !check_gl_error("gl_context::texture_replace_from_buffer", &tb);
    texture_unbind(tb.tt, tmp_id);
 
    if (result && gen_mipmap)
        result = texture_generate_mipmaps(tb, tb.tt == TT_CUBEMAP ? 2 : (int)tb.tt);
 
    return result;
}
 
bool gl_context::texture_copy_back(const texture_base& tb, int level, cgv::data::data_view& dv) const
{
    if(!tb.is_created()) {
        error("gl_context::texture_copy_back: attempt to copy from not created texture", &tb);
        return false;
    }
 
    GLuint tmp_id = texture_bind(tb.tt, get_gl_id(tb.handle));
    read_texture(dv, level);
    
    bool result = !check_gl_error("gl_context::texture_copy_back", &tb);
    texture_unbind(tb.tt, tmp_id);
 
    return result;
}
 
bool gl_context::texture_create_mipmaps(texture_base& tb, cgv::data::data_format& df) const
{
    GLuint gl_format = (const GLuint&)tb.internal_format;
        
    // extract component type
    unsigned int transfer_format = map_to_gl(df.get_standard_component_format(), df.get_integer_interpretation());
 
    if(transfer_format == -1) {
        error("could not determine transfer format", &tb);
        return false;
    }
 
    // extract texture size and compute number of mip-levels
    uvec3 size(unsigned(df.get_width()), unsigned(df.get_height()), unsigned(df.get_depth()));
 
    unsigned max_size = cgv::math::max_value(size);
    unsigned num_levels = 1 + static_cast<unsigned>(log2(static_cast<float>(max_size)));
 
    // compute mip-level sizes
    std::vector<uvec3> level_sizes(num_levels);
    level_sizes[0] = size;
 
    for(unsigned level = 1; level < num_levels; ++level) {
        uvec3 level_size = level_sizes[level - 1];
        level_size = level_size / 2u;
        level_size = cgv::math::max(level_size, uvec3(1u));
        level_sizes[level] = level_size;
    }
 
    GLuint tmp_id = texture_bind(tb.tt, get_gl_id(tb.handle));
 
    bool result = true;
 
    switch(tb.tt) {
    case TT_1D:
        for(unsigned level = 1; level < num_levels; ++level) {
            uvec3 level_size = level_sizes[level];
            glTexImage1D(GL_TEXTURE_1D, level, gl_format, level_size.x(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        }
        break;
    case TT_1D_ARRAY:
        //glTexImage2D(GL_TEXTURE_1D_ARRAY, 0, gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        error("create mipmaps not implemented for 1D array textures", &tb);
        result = false;
        break;
    case TT_2D:
        for(unsigned level = 1; level < num_levels; ++level) {
            uvec3 level_size = level_sizes[level];
            glTexImage2D(GL_TEXTURE_2D, level, gl_format, level_size.x(), level_size.y(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        }
        break;
    case TT_MULTISAMPLE_2D:
        //glTexImage2DMultisample(GL_TEXTURE_2D_MULTISAMPLE, tb.nr_multi_samples, gl_format, df.get_width(), df.get_height(), map_to_gl(tb.fixed_sample_locations));
        error("create mipmaps not implemented for 2D multisample textures", &tb);
        result = false;
        break;
    case TT_2D_ARRAY:
        //glTexImage3D(GL_TEXTURE_2D_ARRAY, 0, gl_format, df.get_width(), df.get_height(), df.get_depth(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        error("create mipmaps not implemented for 2D array textures", &tb);
        result = false;
        break;
    case TT_MULTISAMPLE_2D_ARRAY:
        //glTexStorage3DMultisample(GL_TEXTURE_2D_MULTISAMPLE_ARRAY, tb.nr_multi_samples, gl_format, df.get_width(), df.get_height(), df.get_depth(), map_to_gl(tb.fixed_sample_locations));
        error("create mipmaps not implemented for 2D multisample array textures", &tb);
        result = false;
        break;
    case TT_3D:
        for(unsigned level = 1; level < num_levels; ++level) {
            uvec3 level_size = level_sizes[level];
            glTexImage3D(GL_TEXTURE_3D, level, gl_format, level_size.x(), level_size.y(), level_size.z(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        }
        break;
    case TT_CUBEMAP:
        /*glTexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D(GL_TEXTURE_CUBE_MAP_NEGATIVE_X, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_Y, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D(GL_TEXTURE_CUBE_MAP_NEGATIVE_Y, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_Z, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        glTexImage2D(GL_TEXTURE_CUBE_MAP_NEGATIVE_Z, 0,
                        gl_format, df.get_width(), df.get_height(), 0, transfer_format, GL_UNSIGNED_BYTE, 0);
        */
        error("create mipmaps not implemented for cubemap textures", &tb);
        result = false;
    default:
        break;
    }
 
    if(check_gl_error("gl_context::texture_create_mipmaps", &tb))
        result = false;
 
    if(result)
        tb.have_mipmaps = true;
    
    texture_unbind(tb.tt, tmp_id);
    return result;
}
 
bool gl_context::texture_generate_mipmaps(texture_base& tb, unsigned int dim) const
{
    GLuint tmp_id = texture_bind(tb.tt,get_gl_id(tb.handle));
 
    bool is_array = tb.tt == TT_1D_ARRAY || tb.tt == TT_2D_ARRAY;
    bool is_cubemap = tb.tt == TT_CUBEMAP;
    std::string error_string;
    bool result = generate_mipmaps(dim, is_cubemap, is_array, &error_string);
    if (result)
        tb.have_mipmaps = true;
    else
        error(std::string("gl_context::texture_generate_mipmaps: ") + error_string);
 
    texture_unbind(tb.tt, tmp_id);
    return result;
}
 
bool gl_context::texture_destruct(texture_base& tb) const
{
    if (!tb.is_created()) {
        error("gl_context::texture_destruct: attempt to destruct not created texture", &tb);
        return false;
    }
    GLuint tex_id = get_gl_id(tb.handle);
    glDeleteTextures(1, &tex_id);
    bool result = !check_gl_error("gl_context::texture_destruct", &tb);
    tb.handle = 0;
    tb.ctx_ptr = 0;
    return result;
}
 
bool gl_context::texture_set_state(const texture_base& tb) const
{
    if (tb.tt == TT_UNDEF) {
        error("gl_context::texture_set_state: attempt to set state on texture without type", &tb);
        return false;
    }
    GLuint tex_id = (GLuint&) tb.handle - 1;
    if (tex_id == -1) {
        error("gl_context::texture_set_state: attempt of setting texture state of not created texture", &tb);
        return false;
    }
    GLint tmp_id = texture_bind(tb.tt, tex_id);
 
    if (tb.tt != TT_MULTISAMPLE_2D && tb.tt != TT_MULTISAMPLE_2D_ARRAY) {
        glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_MIN_FILTER, map_to_gl(tb.min_filter));
        glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_MAG_FILTER, map_to_gl(tb.mag_filter));
        if (tb.min_filter == TF_ANISOTROP)
            glTexParameterf(get_tex_dim(tb.tt), GL_TEXTURE_MAX_ANISOTROPY_EXT, tb.anisotropy);
        else
            glTexParameterf(get_tex_dim(tb.tt), GL_TEXTURE_MAX_ANISOTROPY_EXT, 1.0f);
        glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_COMPARE_FUNC, map_to_gl(tb.compare_function));
        glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_COMPARE_MODE, (tb.use_compare_function ? GL_COMPARE_REF_TO_TEXTURE : GL_NONE));
        if (!core_profile)
            glTexParameterf(get_tex_dim(tb.tt), GL_TEXTURE_PRIORITY, tb.priority);
        glTexParameterfv(get_tex_dim(tb.tt), GL_TEXTURE_BORDER_COLOR, tb.border_color);
        //  if (tb.border_color[0] >= 0.0f)
        glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_WRAP_S, map_to_gl(tb.wrap_s));
        if (tb.tt > TT_1D)
            glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_WRAP_T, map_to_gl(tb.wrap_t));
        if (tb.tt == TT_3D)
            glTexParameteri(get_tex_dim(tb.tt), GL_TEXTURE_WRAP_R, map_to_gl(tb.wrap_r));
    }
 
    bool result = !check_gl_error("gl_context::texture_set_state", &tb);
    texture_unbind(tb.tt, tmp_id);
    return result;
}
 
bool gl_context::texture_enable(texture_base& tb, int tex_unit, unsigned int dim) const
{
    if (dim < 1 || dim > 3) {
        error("gl_context::texture_enable: invalid texture dimension", &tb);
        return false;
    }
    GLuint tex_id = (GLuint&) tb.handle - 1;
    if (tex_id == -1) {
        error("gl_context::texture_enable: texture not created", &tb);
        return false;
    }
    if (tex_unit >= 0) {
        if (!GLEW_VERSION_1_3) {
            error("gl_context::texture_enable: multi texturing not supported", &tb);
            return false;
        }
        glActiveTexture(GL_TEXTURE0+tex_unit);
    }
    GLint& old_binding = (GLint&) tb.user_data;
    glGetIntegerv(get_tex_bind(tb.tt), &old_binding);
    ++old_binding;
    glBindTexture(get_tex_dim(tb.tt), tex_id);
    // glEnable is not needed for texture arrays and will throw an invalid enum error
    //if(!(tb.tt == TT_1D_ARRAY || tb.tt == TT_2D_ARRAY))
    //  glEnable(get_tex_dim(tb.tt));
    bool result = !check_gl_error("gl_context::texture_enable", &tb);
    if (tex_unit >= 0)
        glActiveTexture(GL_TEXTURE0);
    return result;
}
 
bool gl_context::texture_disable(
                        texture_base& tb, 
                        int tex_unit, unsigned int dim) const
{
    if (dim < 1 || dim > 3) {
        error("gl_context::texture_disable: invalid texture dimension", &tb);
        return false;
    }
    if (tex_unit == -2) {
        error("gl_context::texture_disable: invalid texture unit", &tb);
        return false;
    }
    GLuint old_binding = (const GLuint&) tb.user_data;
    --old_binding;
    if (tex_unit >= 0)
        glActiveTexture(GL_TEXTURE0+tex_unit);
    // glDisable is not needed for texture arrays and will throw an invalid enum error
    //if(!(tb.tt == TT_1D_ARRAY || tb.tt == TT_2D_ARRAY))
    //  glDisable(get_tex_dim(tb.tt));
    bool result = !check_gl_error("gl_context::texture_disable", &tb);
    glBindTexture(get_tex_dim(tb.tt), old_binding);
    if (tex_unit >= 0)
        glActiveTexture(GL_TEXTURE0);
    return result;
}
 
bool gl_context::texture_bind_as_image(texture_base& tb, int tex_unit, int level, bool bind_array, int layer, AccessType access) const
{   
    GLuint tex_id = (GLuint&)tb.handle - 1;
    if(tex_id == -1) {
        error("gl_context::texture_enable: texture not created", &tb);
        return false;
    }
 
    if(!GLEW_VERSION_4_2) {
        error("gl_context::texture_bind_as_image: image textures not supported", &tb);
        return false;
    }
 
    GLuint gl_format = (const GLuint&)tb.internal_format;
    glBindImageTexture(tex_unit, tex_id, level, map_to_gl(bind_array), layer, map_to_gl(access), gl_format);
    
    bool result = !check_gl_error("gl_context::texture_bind_as_image", &tb);
    return result;
}
 
bool gl_context::render_buffer_create(render_buffer_base& rb, cgv::data::component_format& cf, int& _width, int& _height) const
{
    if (!GLEW_VERSION_3_0) {
        error("gl_context::render_buffer_create: frame buffer objects not supported", &rb);
        return false;
    }
    if (_width == -1)
        _width = get_width();
    if (_height == -1)
        _height = get_height();
 
    GLuint rb_id;
    glGenRenderbuffers(1, &rb_id);
    glBindRenderbuffer(GL_RENDERBUFFER, rb_id);
 
    GLuint& gl_format = (GLuint&)rb.internal_format;
    unsigned i = find_best_match(cf, color_buffer_formats);
    cgv::data::component_format best_cf(color_buffer_formats[i]);
    gl_format = gl_color_buffer_format_ids[i];
 
    i = find_best_match(cf, depth_formats, &best_cf);
    if (i != -1) {
        best_cf = cgv::data::component_format(depth_formats[i]);
        gl_format = gl_depth_format_ids[i];
    }
 
    cf = best_cf;
    if (rb.nr_multi_samples == 0)
        glRenderbufferStorage(GL_RENDERBUFFER, gl_format, _width, _height);
    else
        glRenderbufferStorageMultisample(GL_RENDERBUFFER, rb.nr_multi_samples, gl_format, _width, _height);
 
    if (check_gl_error("gl_context::render_buffer_create", &rb))
        return false;
    rb.handle = get_handle(rb_id);
    return true;
}
 
bool gl_context::render_buffer_destruct(render_buffer_base& rc) const
{
    if (!GLEW_VERSION_3_0) {
        error("gl_context::render_buffer_destruct: frame buffer objects not supported", &rc);
        return false;
    }
    GLuint rb_id = ((GLuint&) rc.handle)+1;
    glDeleteRenderbuffers(1, &rb_id);
    if (check_gl_error("gl_context::render_buffer_destruct", &rc))
        return false;
    rc.handle = 0;
    rc.ctx_ptr = 0;
    return true;
}
 
bool gl_context::frame_buffer_create(frame_buffer_base& fbb) const
{
    if (!check_fbo_support("gl_context::frame_buffer_create", &fbb))
        return false;
    
    if (!context::frame_buffer_create(fbb))
        return false;
 
    // allocate framebuffer object
    GLuint fbo_id = 0;
    glGenFramebuffers(1, &fbo_id);
    if (fbo_id == 0) {
        error("gl_context::frame_buffer_create: could not allocate framebuffer object", &fbb);
        return false;
    }
    fbb.handle = get_handle(fbo_id);
    return true;
}
 
bool gl_context::frame_buffer_enable(frame_buffer_base& fbb)
{
    if (!context::frame_buffer_enable(fbb))
        return false;
    std::vector<int> buffers;
    bool depth_buffer = false;
    get_buffer_list(fbb, depth_buffer, buffers, GL_COLOR_ATTACHMENT0);
    glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(fbb.handle));
 
    if (buffers.size() == 1)
        glDrawBuffer(buffers[0]);
    else if (buffers.size() > 1) {
        glDrawBuffers(GLsizei(buffers.size()), reinterpret_cast<GLenum*>(&buffers[0]));
    }
    else if(depth_buffer) {
        glDrawBuffer(GL_NONE);
        //glReadBuffer(GL_NONE);
    } else {
        error("gl_context::frame_buffer_enable: no attached draw buffer selected!!", &fbb);
        return false;
    }
    return true;
}

bool gl_context::frame_buffer_disable(frame_buffer_base& fbb)
{
    if (!context::frame_buffer_disable(fbb))
        return false;
    if (frame_buffer_stack.empty()) {
        error("gl_context::frame_buffer_disable called with empty frame buffer stack!!", &fbb);
        return false;
    }
    else
        glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(frame_buffer_stack.top()->handle));
    return true;
}
 
bool gl_context::frame_buffer_destruct(frame_buffer_base& fbb) const
{
    if (!context::frame_buffer_destruct(fbb))
        return false;
    GLuint fbo_id = get_gl_id(fbb.handle);
    glDeleteFramebuffers(1, &fbo_id);
    fbb.handle = 0;
    fbb.ctx_ptr = 0;
    return true;
}
 
void complete_rect_from_vp(ivec4& D, GLint vp[4])
{
    if (D(0) == -1)
        D(0) = vp[0];
    if (D(1) == -1)
        D(1) = vp[1];
    if (D(2) == -1)
        D(2) = vp[0] + vp[2];
    if (D(3) == -1)
        D(3) = vp[1] + vp[3];
}
 
void gl_context::frame_buffer_blit(
    const frame_buffer_base* src_fbb_ptr, const ivec4& _S,
    frame_buffer_base* dst_fbb_ptr, const ivec4& _D, BufferTypeBits btbs, bool interpolate) const
{
    static const GLenum masks[8]{
        0,
        GL_COLOR_BUFFER_BIT,
        GL_DEPTH_BUFFER_BIT,
        GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT,
        GL_STENCIL_BUFFER_BIT,
        GL_STENCIL_BUFFER_BIT | GL_COLOR_BUFFER_BIT,
        GL_STENCIL_BUFFER_BIT | GL_DEPTH_BUFFER_BIT,
        GL_STENCIL_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_COLOR_BUFFER_BIT
    };
    ivec4 S = _S;
    ivec4 D = _D;
    if ((src_fbb_ptr == 0 && (S(0) == -1 || S(1) == -1 || S(2) == -1 || S(3) == -1)) ||
        (dst_fbb_ptr == 0 && (D(0) == -1 || D(1) == -1 || D(2) == -1 || D(3) == -1))) {
        GLint vp[4];
        glGetIntegerv(GL_VIEWPORT, vp);
        if (src_fbb_ptr == 0)
            complete_rect_from_vp(S, vp);
        if (dst_fbb_ptr == 0)
            complete_rect_from_vp(D, vp);
    }
    GLint old_draw_fbo, old_read_fbo;
    if (src_fbb_ptr) {
        glGetIntegerv(GL_READ_FRAMEBUFFER_BINDING, &old_read_fbo);
        glBindFramebuffer(GL_READ_FRAMEBUFFER, get_gl_id(src_fbb_ptr->handle));
    }
    if (dst_fbb_ptr) {
        glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &old_draw_fbo);
        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, get_gl_id(dst_fbb_ptr->handle));
    }
    glBlitFramebuffer(S(0), S(1), S(2), S(3), D(0), D(1), D(2), D(3), masks[btbs], interpolate ? GL_LINEAR : GL_NEAREST);
    if (src_fbb_ptr) 
        glBindFramebuffer(GL_READ_FRAMEBUFFER, old_read_fbo);
    if (dst_fbb_ptr)
        glBindFramebuffer(GL_DRAW_FRAMEBUFFER, old_draw_fbo);
}
 
bool gl_context::frame_buffer_attach(frame_buffer_base& fbb, const render_buffer_base& rb, bool is_depth, int i) const
{
    if (!context::frame_buffer_attach(fbb, rb, is_depth, i))
        return false;
    GLint old_binding;
    glGetIntegerv(GL_FRAMEBUFFER_BINDING, &old_binding);
    glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(fbb.handle));
        glFramebufferRenderbuffer(GL_FRAMEBUFFER, 
            is_depth ? GL_DEPTH_ATTACHMENT : GL_COLOR_ATTACHMENT0 + i,
            GL_RENDERBUFFER, 
            get_gl_id(rb.handle));
    glBindFramebuffer(GL_FRAMEBUFFER, old_binding);
    return true;
}

bool gl_context::frame_buffer_attach(frame_buffer_base& fbb, 
                                                 const texture_base& t, bool is_depth,
                                                 int level, int i, int z_or_cube_side) const
{
    if (!context::frame_buffer_attach(fbb, t, is_depth, level, i, z_or_cube_side))
        return false;
    GLint old_binding;
    glGetIntegerv(GL_FRAMEBUFFER_BINDING, &old_binding);
    glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(fbb.handle));
 
    if (z_or_cube_side == -1) {
        glFramebufferTexture2D(GL_FRAMEBUFFER, 
            is_depth ? GL_DEPTH_ATTACHMENT : GL_COLOR_ATTACHMENT0+i,
            t.tt == TT_2D ? GL_TEXTURE_2D : GL_TEXTURE_2D_MULTISAMPLE, get_gl_id(t.handle), level);
    }
    else {
        if (t.tt == TT_CUBEMAP) {
            glFramebufferTexture2D(GL_FRAMEBUFFER,
                GL_COLOR_ATTACHMENT0+i,
                get_gl_cube_map_target(z_or_cube_side), get_gl_id(t.handle), level);
        }
        else {
            glFramebufferTexture3D(GL_FRAMEBUFFER, 
                GL_COLOR_ATTACHMENT0+i, 
                t.tt == TT_3D ? GL_TEXTURE_3D : GL_TEXTURE_2D_MULTISAMPLE_ARRAY , get_gl_id(t.handle), level, z_or_cube_side);
        }
    }
    bool result = !check_gl_error("gl_context::frame_buffer_attach", &fbb);
    glBindFramebuffer(GL_FRAMEBUFFER, old_binding);
    return result;
}

bool gl_context::frame_buffer_is_complete(const frame_buffer_base& fbb) const
{
    if (fbb.handle == 0) {
        error("gl_context::frame_buffer_is_complete: attempt to check completeness on frame buffer that is not created", &fbb);
        return false;
    }
    GLint old_binding;
    glGetIntegerv(GL_FRAMEBUFFER_BINDING, &old_binding);
    glBindFramebuffer(GL_FRAMEBUFFER, get_gl_id(fbb.handle));
    GLenum error = glCheckFramebufferStatus(GL_FRAMEBUFFER);
    glBindFramebuffer(GL_FRAMEBUFFER, old_binding);
    switch (error) {
    case GL_FRAMEBUFFER_COMPLETE:
        return true;
    case GL_FRAMEBUFFER_UNDEFINED:
        fbb.last_error = "undefined framebuffer";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT:
        fbb.last_error = "incomplete attachment";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_MISSING_ATTACHMENT:
      fbb.last_error = "incomplete or missing attachment";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_MULTISAMPLE:
      fbb.last_error = "incomplete multisample";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS:
      fbb.last_error = "incomplete layer targets";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_DRAW_BUFFER:
      fbb.last_error = "incomplete draw buffer";
        return false;
    case GL_FRAMEBUFFER_INCOMPLETE_READ_BUFFER:
      fbb.last_error = "incomplete read buffer";
        return false;
    case GL_FRAMEBUFFER_UNSUPPORTED:
      fbb.last_error = "framebuffer objects unsupported";
        return false;
    }
    fbb.last_error = "unknown error";
    return false;
}
 
int gl_context::frame_buffer_get_max_nr_color_attachments() const
{
    if (!check_fbo_support("gl_context::frame_buffer_get_max_nr_color_attachments"))
        return 0;
 
    GLint nr;
    glGetIntegerv(GL_MAX_COLOR_ATTACHMENTS, &nr);
    return nr;
}
 
int gl_context::frame_buffer_get_max_nr_draw_buffers() const
{
    if (!check_fbo_support("gl_context::frame_buffer_get_max_nr_draw_buffers"))
        return 0;
 
    GLint nr;
    glGetIntegerv(GL_MAX_DRAW_BUFFERS, &nr);
    return nr;
}
 
GLuint gl_shader_type[] = 
{
    0, GL_COMPUTE_SHADER, GL_VERTEX_SHADER, GL_TESS_CONTROL_SHADER, GL_TESS_EVALUATION_SHADER, GL_GEOMETRY_SHADER, GL_FRAGMENT_SHADER, GL_TASK_SHADER_EXT, GL_MESH_SHADER_EXT
};
 
void gl_context::shader_code_destruct(render_component& sc) const
{
    if (sc.handle == 0) {
        error("gl_context::shader_code_destruct: shader not created", &sc);
        return;
    }
    glDeleteShader(get_gl_id(sc.handle));
    sc.handle = 0;
    sc.ctx_ptr = 0;
    check_gl_error("gl_context::shader_code_destruct", &sc);
}
 
bool gl_context::shader_code_create(render_component& sc, ShaderType st, const std::string& source) const
{
    if (!check_shader_support(st, "gl_context::shader_code_create", &sc))
        return false;
 
    while (glGetError()); // Clear error flags.
 
    GLuint s_id = glCreateShader(gl_shader_type[st]);
    if (s_id == -1) {
        error(std::string("gl_context::shader_code_create: ")+gl_error(), &sc);
        return false;
    }
    sc.handle = get_handle(s_id);
 
    const char* s = source.c_str();
    glShaderSource(s_id, 1, &s,NULL);
    if (check_gl_error("gl_context::shader_code_create", &sc))
        return false;
 
    return true;
}
 
bool gl_context::shader_code_compile(render_component& sc) const
{
    if (sc.handle == 0) {
        error("gl_context::shader_code_compile: shader not created", &sc);
        return false;
    }
    GLuint s_id = get_gl_id(sc.handle);
    glCompileShader(s_id);
    GLint result = GL_FALSE;
    glGetShaderiv(s_id, GL_COMPILE_STATUS, &result);
    if (result == GL_TRUE)
        return true;
    sc.last_error = std::string();
    GLint infologLength = 0;
    glGetShaderiv(s_id, GL_INFO_LOG_LENGTH, &infologLength);
    if (infologLength > 0) {
        GLsizei charsWritten = 0;
        sc.last_error = std::string(infologLength, 0);
        glGetShaderInfoLog(s_id, infologLength, &charsWritten, &sc.last_error.front());
        sc.last_error.resize(static_cast<size_t>(charsWritten));
    }
    return false;
}
 
bool gl_context::shader_program_create(shader_program_base& spb) const
{
    if (!check_shader_support(ST_VERTEX, "gl_context::shader_program_create", &spb))
        return false;
    spb.handle = get_handle(glCreateProgram());
    return true;
}
 
void gl_context::shader_program_attach(shader_program_base& spb, const render_component& sc) const
{
    if (spb.handle == 0) {
        error("gl_context::shader_program_attach: shader program not created", &spb);
        return;
    }
    glAttachShader(get_gl_id(spb.handle), get_gl_id(sc.handle));
}
 
void gl_context::shader_program_detach(shader_program_base& spb, const render_component& sc) const
{
    if (spb.handle == 0) {
        error("gl_context::shader_program_detach: shader program not created", &spb);
        return;
    }
    glDetachShader(get_gl_id(spb.handle), get_gl_id(sc.handle));
}
 
bool gl_context::shader_program_link(shader_program_base& spb) const
{
    if (spb.handle == 0) {
        error("gl_context::shader_program_link: shader program not created", &spb);
        return false;
    }
    GLuint p_id = get_gl_id(spb.handle);
    glLinkProgram(p_id); 
    GLint result = GL_FALSE;
    glGetProgramiv(p_id, GL_LINK_STATUS, &result);
    if (result == GL_TRUE)
        return context::shader_program_link(spb);
    GLint infologLength = 0;
    glGetProgramiv(p_id, GL_INFO_LOG_LENGTH, &infologLength);
    if (infologLength > 0) {
        GLsizei charsWritten = 0;
        spb.last_error = std::string(infologLength, 0);
        glGetProgramInfoLog(p_id, infologLength, &charsWritten, &spb.last_error.front());
        spb.last_error.resize(static_cast<size_t>(charsWritten));
        error("gl_context::shader_program_link\n" + spb.last_error, &spb);
    }
 
    return false;
}
 
bool gl_context::shader_program_set_state(shader_program_base& spb) const
{
    if (spb.handle == 0) {
        error("gl_context::shader_program_set_state: shader program not created", &spb);
        return false;
    }
    GLuint p_id = get_gl_id(spb.handle);
    glProgramParameteri(p_id, GL_GEOMETRY_VERTICES_OUT_ARB, spb.geometry_shader_output_count);
    glProgramParameteri(p_id, GL_GEOMETRY_INPUT_TYPE_ARB, map_to_gl(spb.geometry_shader_input_type));
    glProgramParameteri(p_id, GL_GEOMETRY_OUTPUT_TYPE_ARB, map_to_gl(spb.geometry_shader_output_type));
    return true;
}
 
bool gl_context::shader_program_enable(shader_program_base& spb)
{
    if (!context::shader_program_enable(spb))
        return false;
    glUseProgram(get_gl_id(spb.handle));    
    shader_program& prog = static_cast<shader_program&>(spb);
    if (auto_set_lights_in_current_shader_program && spb.does_use_lights())
        set_current_lights(prog);
    if (auto_set_material_in_current_shader_program && spb.does_use_material())
        set_current_material(prog);
    if (auto_set_view_in_current_shader_program && spb.does_use_view())
        set_current_view(prog);
    if (auto_set_gamma_in_current_shader_program && spb.does_use_gamma())
        set_current_gamma(prog);
    if (prog.does_context_set_color() && prog.get_color_index() >= 0)
        prog.set_attribute(*this, prog.get_color_index(), current_color);
    return true;
}
 
bool gl_context::shader_program_disable(shader_program_base& spb)
{
    if (!context::shader_program_disable(spb)) 
        return false;
    if (shader_program_stack.empty())
        glUseProgram(0);
    else
        glUseProgram(get_gl_id(shader_program_stack.top()->handle));
    return true;
}
 
bool gl_context::shader_program_destruct(shader_program_base& spb) const
{
    if (!context::shader_program_destruct(spb))
        return false;
    glDeleteProgram(get_gl_id(spb.handle));
    spb.handle = 0;
    spb.ctx_ptr = 0;
    return true;
}
 
bool gl_context::shader_program_get_active_uniforms(shader_program_base& spb, std::vector<std::string>& names) const
{
    if (spb.handle == 0)
        return false;
 
    GLuint p_id = get_gl_id(spb.handle);
 
    GLint num_active_uniforms = 0;
    glGetProgramiv(p_id, GL_ACTIVE_UNIFORMS, &num_active_uniforms);
 
    names.reserve(num_active_uniforms);
 
    std::vector<GLchar> buffer(256);
    for (int i = 0; i < num_active_uniforms; ++i) {
        GLint array_size = 0;
        GLenum type = 0;
        GLsizei actual_length = 0;
        
        glGetActiveUniform(p_id, i, GLsizei(buffer.size()), &actual_length, &array_size, &type, buffer.data());
        std::string name(static_cast<char*>(buffer.data()), actual_length);
 
        // Uniforms for arrays of non-compound (non-struct) types are listed once with a "[0]" suffix and a given array size greater than 1.
        if(array_size > 1) {
            // Remove the brackets to get the base name of the uniform
            size_t bracket_pos = name.find('[');
            if(bracket_pos != std::string::npos)
                name.resize(bracket_pos);
 
            if(!name.empty()) {
                // Store the name without the brackets to allow setting complete arrays using just the uniform name.
                names.push_back(name);
                // Additionally store an entry for every possible indexed name to allow setting elements individually.
                for(GLint i = 0; i < array_size; ++i)
                    names.push_back(name + "[" + std::to_string(i) + "]");
            }
        } else {
            if(!name.empty())
                names.push_back(name);
        }
    }
    return true;
}
 
program_variable_info construct_program_variable(const std::vector<GLchar>& buffer, GLsizei actual_length, GLint array_size, GLenum type)
{
    program_variable_info V;
    V.name = std::string(static_cast<const char*>(buffer.data()), actual_length);
    // Uniforms for arrays of non-compound (non-struct) types are listed once with a "[0]" suffix and a given array size greater than 1.
    if (array_size > 1) {
        // Remove the brackets to get the base name of the uniform
        size_t bracket_pos = V.name.find('[');
        if (bracket_pos != std::string::npos)
            V.name.resize(bracket_pos);
        if (!V.name.empty())
            V.array_size = array_size;
    }
    else {
        V.array_size = 1;
    }
    switch (type) {
    case GL_FLOAT: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, false); break;
    case GL_FLOAT_VEC2: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 2u); break;
    case GL_FLOAT_VEC3: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 3u); break;
    case GL_FLOAT_VEC4: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 4u); break;
    case GL_FLOAT_MAT2: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 2u, 2u, false); break;
    case GL_FLOAT_MAT3: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 3u, 3u, false); break;
    case GL_FLOAT_MAT4: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 4u, 4u, false); break;
    case GL_FLOAT_MAT2x3: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 2u, 3u, false); break;
    case GL_FLOAT_MAT2x4: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 2u, 4u, false); break;
    case GL_FLOAT_MAT3x2: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 3u, 2u, false); break;
    case GL_FLOAT_MAT3x4: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 3u, 4u, false); break;
    case GL_FLOAT_MAT4x2: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 4u, 2u, false); break;
    case GL_FLOAT_MAT4x3: V.type_descr = type_descriptor(cgv::type::info::TI_FLT32, 4u, 3u, false); break;
    case GL_BOOL: V.type_descr = type_descriptor(cgv::type::info::TI_BOOL, false); break;
    case GL_BOOL_VEC2:V.type_descr = type_descriptor(cgv::type::info::TI_BOOL, 2u); break;
    case GL_BOOL_VEC3:V.type_descr = type_descriptor(cgv::type::info::TI_BOOL, 3u); break;
    case GL_BOOL_VEC4:V.type_descr = type_descriptor(cgv::type::info::TI_BOOL, 4u); break;
    case GL_INT: V.type_descr = type_descriptor(cgv::type::info::TI_INT32, false); break;
    case GL_INT_VEC2:V.type_descr = type_descriptor(cgv::type::info::TI_INT32, 2u); break;
    case GL_INT_VEC3:V.type_descr = type_descriptor(cgv::type::info::TI_INT32, 3u); break;
    case GL_INT_VEC4:V.type_descr = type_descriptor(cgv::type::info::TI_INT32, 4u); break;
    case GL_UNSIGNED_INT:V.type_descr = type_descriptor(cgv::type::info::TI_UINT32, false); break;
    case GL_UNSIGNED_INT_VEC2:V.type_descr = type_descriptor(cgv::type::info::TI_UINT32, 2u); break;
    case GL_UNSIGNED_INT_VEC3:V.type_descr = type_descriptor(cgv::type::info::TI_UINT32, 3u); break;
    case GL_UNSIGNED_INT_VEC4:V.type_descr = type_descriptor(cgv::type::info::TI_UINT32, 4u); break;
    case GL_DOUBLE: V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, false); break;
    case GL_DOUBLE_VEC2:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 2u); break;
    case GL_DOUBLE_VEC3:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 3u); break;
    case GL_DOUBLE_VEC4:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 4u); break;
    case GL_DOUBLE_MAT2:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 2u, 2u, false); break;
    case GL_DOUBLE_MAT3:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 3u, 3u, false); break;
    case GL_DOUBLE_MAT4:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 4u, 4u, false); break;
    case GL_DOUBLE_MAT2x3:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 2u, 3u, false); break;
    case GL_DOUBLE_MAT2x4:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 2u, 4u, false); break;
    case GL_DOUBLE_MAT3x2:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 3u, 2u, false); break;
    case GL_DOUBLE_MAT3x4:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 3u, 4u, false); break;
    case GL_DOUBLE_MAT4x2:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 4u, 2u, false); break;
    case GL_DOUBLE_MAT4x3:V.type_descr = type_descriptor(cgv::type::info::TI_FLT64, 4u, 3u, false); break;
    }
    return V;
}
 
void gl_context::shader_program_inspect_variables(shader_program_base& spb, ProgramVariableKind kind, std::vector<program_variable_info>& Vs, bool get_location, bool get_value) const
{
    GLint prog_id;
    spb.put_id(prog_id);
    // name buffer and sizeing informations
    std::vector<GLchar> buffer(256);
    GLint array_size = 0;
    GLenum type = 0;
    GLsizei actual_length = 0;
    // query number of active program variables of \c kind
    GLint num_active_variables = 0;
    glGetProgramiv(prog_id, kind == cgv::render::PVK_UNIFORM ? GL_ACTIVE_UNIFORMS : GL_ACTIVE_ATTRIBUTES, &num_active_variables);
    // for inspecting attribute values, one has to use the to be inspected program. Remember previously used program
    GLint old_prog_id = prog_id;
    if (kind == cgv::render::PVK_ATTRIBUTE) {
        glGetIntegerv(GL_CURRENT_PROGRAM, &old_prog_id);
        if (old_prog_id != prog_id)
            glUseProgram(prog_id);
    }
    // iterate all variables of \c kind
    for (int i = 0; i < num_active_variables; ++i) {
        // extract type and array information of program variable
        (kind == cgv::render::PVK_UNIFORM ? glGetActiveUniform : glGetActiveAttrib)(prog_id, i,
            GLsizei(buffer.size()), &actual_length, &array_size, &type, buffer.data());
        cgv::render::program_variable_info V = construct_program_variable(buffer, actual_length, array_size, type);
        // compute component count \c cnt, array element size \c s and total size \c S for all values in array
        size_t cnt = 1, s, S;
        if (get_value) {
            V.compute_sizes(cnt, s, S);
            V.current_value.resize(S);
        }
        // in case of attributes provide a read buffer to read array elements which write 
        // always 4 components even if only one component is declared
        std::vector<char> read_buffer;
        if (kind == cgv::render::PVK_ATTRIBUTE && cnt != 4)
            read_buffer.resize(4 * s / cnt);
        for (unsigned j = 0; j < V.array_size; ++j) {
            // in case of array variables use per array element postfix to query location
            std::string postfix;
            if (V.array_size > 1)
                postfix += "[" + cgv::utils::to_string(j) + "]";
            // query location of variable in 
            if (get_location || get_value)
                V.program_location = (kind == cgv::render::PVK_UNIFORM ? glGetUniformLocation : glGetAttribLocation)(
                    prog_id, (V.name + postfix).c_str());
            // one is not allowed to query program attribute at location 0
            if (kind == cgv::render::PVK_ATTRIBUTE && V.program_location == 0) {
                V.current_value.clear();
                break;
            }
            if (!get_value)
                break;
            // let read_ptr point where program variable values should be queried to
            void* value_ptr = V.current_value.data() + j * s;
            void* read_ptr = read_buffer.empty() ? value_ptr : read_buffer.data();
            // switch over coordinate type to vary gl getter function
            switch (V.type_descr.coordinate_type) {
            case cgv::type::info::TI_BOOL:
            case cgv::type::info::TI_INT32:
                if (kind == cgv::render::PVK_UNIFORM)
                    glGetUniformiv(prog_id, V.program_location, (GLint*)(read_ptr));
                else
                    glGetVertexAttribIiv(V.program_location, GL_CURRENT_VERTEX_ATTRIB, (GLint*)(read_ptr));
                break;
            case cgv::type::info::TI_UINT32:
                if (kind == cgv::render::PVK_UNIFORM)
                    glGetUniformuiv(prog_id, V.program_location, (GLuint*)(read_ptr));
                else
                    glGetVertexAttribIuiv(V.program_location, GL_CURRENT_VERTEX_ATTRIB, (GLuint*)(read_ptr));
                break;
            case cgv::type::info::TI_FLT32:
                if (kind == cgv::render::PVK_UNIFORM)
                    glGetUniformfv(prog_id, V.program_location, (GLfloat*)(read_ptr));
                else
                    glGetVertexAttribfv(V.program_location, GL_CURRENT_VERTEX_ATTRIB, (GLfloat*)(read_ptr));
                break;
            case cgv::type::info::TI_FLT64:
                if (kind == cgv::render::PVK_UNIFORM)
                    glGetUniformdv(prog_id, V.program_location, (GLdouble*)(read_ptr));
                else
                    glGetVertexAttribdv(V.program_location, GL_CURRENT_VERTEX_ATTRIB, (GLdouble*)(read_ptr));
                break;
            }
            // in case we copied to read buffer, copy used value[s] to value buffer
            if (read_ptr != value_ptr)
                std::copy((const char*)read_ptr, (const char*)read_ptr + s, (char*)value_ptr);
        }
        // add inspected variable to list
        Vs.push_back(V);
    }
    // recover previously used program
    if (old_prog_id != prog_id)
        glUseProgram(old_prog_id);
}
 
 
int  gl_context::get_uniform_location(const shader_program_base& spb, const std::string& name) const
{
    return glGetUniformLocation(get_gl_id(spb.handle), name.c_str());
}
 
std::string value_type_index_to_string(type_descriptor td)
{
    std::string res = cgv::type::info::get_type_name(td.coordinate_type);
    switch (td.element_type) {
    case ET_VECTOR:
        res = std::string("vector<") + res + "," + cgv::utils::to_string(td.nr_rows) + ">";
        break;
    case ET_MATRIX:
        res = std::string("matrix<") + res + "," + cgv::utils::to_string(td.nr_rows) + "," + cgv::utils::to_string(td.nr_columns) + ">";
        if (td.is_row_major)
            res += "^T";
    default:
        break;
    }
    if (td.is_array)
        res += "[]";
    return res;
}
 
bool gl_context::set_uniform_void(shader_program_base& spb, int loc, type_descriptor value_type, const void* value_ptr) const
{
    if (value_type.is_array) {
        error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") array type not supported, please use set_uniform_array instead.", &spb);
        return false;
    }
    if (!spb.handle) {
        error("gl_context::set_uniform_void() called on not created program", &spb);
        return false;
    }
    bool not_current = shader_program_stack.empty() || shader_program_stack.top() != &spb;
    if (not_current)
        glUseProgram(get_gl_id(spb.handle));
    bool res = true;
    switch (value_type.element_type) {
    case ET_VALUE:
        switch (value_type.coordinate_type) {
        case TI_BOOL: glUniform1i(loc, *reinterpret_cast<const bool*>(value_ptr) ? 1 : 0); break;
        case TI_UINT8: glUniform1ui(loc, *reinterpret_cast<const uint8_type*>(value_ptr)); break;
        case TI_UINT16: glUniform1ui(loc, *reinterpret_cast<const uint16_type*>(value_ptr)); break;
        case TI_UINT32: glUniform1ui(loc, *reinterpret_cast<const uint32_type*>(value_ptr)); break;
        case TI_INT8: glUniform1i(loc, *reinterpret_cast<const int8_type*>(value_ptr)); break;
        case TI_INT16: glUniform1i(loc, *reinterpret_cast<const int16_type*>(value_ptr)); break;
        case TI_INT32: glUniform1i(loc, *reinterpret_cast<const int32_type*>(value_ptr)); break;
        case TI_FLT32: glUniform1f(loc, *reinterpret_cast<const flt32_type*>(value_ptr)); break;
        case TI_FLT64: glUniform1d(loc, *reinterpret_cast<const flt64_type*>(value_ptr)); break;
        default:
            error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
            res = false; break;
        }
        break;
    case ET_VECTOR:
        switch (value_type.nr_rows) {
        case 2:
            switch (value_type.coordinate_type) {
            case TI_BOOL:   glUniform2i(loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0); break;
            case TI_UINT8:  glUniform2ui(loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1]); break;
            case TI_UINT16: glUniform2ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1]); break;
            case TI_UINT32: glUniform2ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1]); break;
            case TI_INT8:   glUniform2i(loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1]); break;
            case TI_INT16:  glUniform2i(loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1]); break;
            case TI_INT32:  glUniform2i(loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1]); break;
            case TI_FLT32:  glUniform2f(loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1]); break;
            case TI_FLT64:  glUniform2d(loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1]); break;
            default:
                error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false; break;
            }
            break;
        case 3:
            switch (value_type.coordinate_type) {
            case TI_BOOL:   glUniform3i(loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[2] ? 1 : 0); break;
            case TI_UINT8:  glUniform3ui(loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1], reinterpret_cast<const uint8_type*> (value_ptr)[2]); break;
            case TI_UINT16: glUniform3ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1], reinterpret_cast<const uint16_type*>(value_ptr)[2]); break;
            case TI_UINT32: glUniform3ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1], reinterpret_cast<const uint32_type*>(value_ptr)[2]); break;
            case TI_INT8:   glUniform3i(loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1], reinterpret_cast<const int8_type*>  (value_ptr)[2]); break;
            case TI_INT16:  glUniform3i(loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1], reinterpret_cast<const int16_type*> (value_ptr)[2]); break;
            case TI_INT32:  glUniform3i(loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1], reinterpret_cast<const int32_type*> (value_ptr)[2]); break;
            case TI_FLT32:  glUniform3f(loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1], reinterpret_cast<const flt32_type*> (value_ptr)[2]); break;
            case TI_FLT64:  glUniform3d(loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1], reinterpret_cast<const flt64_type*> (value_ptr)[2]); break;
            default:
                error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false; break;
            }
            break;
        case 4:
            switch (value_type.coordinate_type) {
            case TI_BOOL:   glUniform4i(loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[2] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[3] ? 1 : 0); break;
            case TI_UINT8:  glUniform4ui(loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1], reinterpret_cast<const uint8_type*> (value_ptr)[2], reinterpret_cast<const uint8_type*> (value_ptr)[3]); break;
            case TI_UINT16: glUniform4ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1], reinterpret_cast<const uint16_type*>(value_ptr)[2], reinterpret_cast<const uint16_type*>(value_ptr)[3]); break;
            case TI_UINT32: glUniform4ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1], reinterpret_cast<const uint32_type*>(value_ptr)[2], reinterpret_cast<const uint32_type*>(value_ptr)[3]); break;
            case TI_INT8:   glUniform4i(loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1], reinterpret_cast<const int8_type*>  (value_ptr)[2], reinterpret_cast<const int8_type*>  (value_ptr)[3]); break;
            case TI_INT16:  glUniform4i(loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1], reinterpret_cast<const int16_type*> (value_ptr)[2], reinterpret_cast<const int16_type*> (value_ptr)[3]); break;
            case TI_INT32:  glUniform4i(loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1], reinterpret_cast<const int32_type*> (value_ptr)[2], reinterpret_cast<const int32_type*> (value_ptr)[3]); break;
            case TI_FLT32:  glUniform4f(loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1], reinterpret_cast<const flt32_type*> (value_ptr)[2], reinterpret_cast<const flt32_type*> (value_ptr)[3]); break;
            case TI_FLT64:  glUniform4d(loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1], reinterpret_cast<const flt64_type*> (value_ptr)[2], reinterpret_cast<const flt64_type*> (value_ptr)[3]); break;
            default:
                error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false; break;
            }
            break;
        }
        break;
    case ET_MATRIX:
        switch (value_type.coordinate_type) {
        case TI_FLT32:
            switch (value_type.nr_rows) {
            case 2:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix2fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 3: glUniformMatrix2x3fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 4: glUniformMatrix2x4fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            case 3:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix3x2fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 3: glUniformMatrix3fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 4: glUniformMatrix3x4fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            case 4:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix4x2fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 3: glUniformMatrix4x3fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                case 4: glUniformMatrix4fv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt32_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            default:
                error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of rows outside [2,..,4].", &spb);
                res = false; break;
            }
            break;
        case TI_FLT64:
            switch (value_type.nr_rows) {
            case 2:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix2dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 3: glUniformMatrix2x3dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 4: glUniformMatrix2x4dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            case 3:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix3x2dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 3: glUniformMatrix3dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 4: glUniformMatrix3x4dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            case 4:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix4x2dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 3: glUniformMatrix4x3dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                case 4: glUniformMatrix4dv(loc, 1, !value_type.is_row_major, reinterpret_cast<const flt64_type*> (value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..,4].", &spb);
                    res = false; break;
                }
                break;
            default:
                error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") matrix number of rows outside [2,..,4].", &spb);
                res = false; break;
            }
            break;
        default:
            error(std::string("gl_context::set_uniform_void(") + value_type_index_to_string(value_type) + ") non float coordinate type not supported.", &spb);
            res = false; break;
        }
        break;
    }
    if (not_current)
        glUseProgram(shader_program_stack.empty() ? 0 : get_gl_id(shader_program_stack.top()->handle));
 
    if (check_gl_error("gl_context::set_uniform_void()", &spb))
        res = false;
    return res;
}
 
bool gl_context::set_uniform_array_void(shader_program_base& spb, int loc, type_descriptor value_type, const void* value_ptr, size_t nr_elements) const
{
    if (!value_type.is_array) {
        error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") non array type not allowed.", &spb);
        return false;
    }
    if (!spb.handle) {
        error("gl_context::set_uniform_array_void() called on not created program", &spb);
        return false;
    }
    bool not_current = shader_program_stack.empty() || shader_program_stack.top() != &spb;
    if (not_current)
        glUseProgram(get_gl_id(spb.handle));
    bool res = true;
    switch (value_type.coordinate_type) {
    case TI_INT32:
        switch (value_type.element_type) {
        case ET_VALUE:
            glUniform1iv(loc, GLsizei(nr_elements), reinterpret_cast<const int32_type*>(value_ptr));
            break;
        case ET_VECTOR:
            switch (value_type.nr_rows) {
            case 2: glUniform2iv(loc, GLsizei(nr_elements), reinterpret_cast<const int32_type*>(value_ptr)); break;
            case 3: glUniform3iv(loc, GLsizei(nr_elements), reinterpret_cast<const int32_type*>(value_ptr)); break;
            case 4: glUniform4iv(loc, GLsizei(nr_elements), reinterpret_cast<const int32_type*>(value_ptr)); break;
            default:
                error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") vector dimension outside [2,..4].", &spb);
                res = false;
                break;
            }
            break;
        case ET_MATRIX:
            error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") type not supported.", &spb);
            res = false;
            break;
        }
        break;
    case TI_UINT32:
        switch (value_type.element_type) {
        case ET_VALUE:
            glUniform1uiv(loc, GLsizei(nr_elements), reinterpret_cast<const uint32_type*>(value_ptr));
            break;
        case ET_VECTOR:
            switch (value_type.nr_rows) {
            case 2: glUniform2uiv(loc, GLsizei(nr_elements), reinterpret_cast<const uint32_type*>(value_ptr)); break;
            case 3: glUniform3uiv(loc, GLsizei(nr_elements), reinterpret_cast<const uint32_type*>(value_ptr)); break;
            case 4: glUniform4uiv(loc, GLsizei(nr_elements), reinterpret_cast<const uint32_type*>(value_ptr)); break;
            default:
                error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") vector dimension outside [2,..4].", &spb);
                res = false;
                break;
            }
            break;
        case ET_MATRIX:
            error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") type not supported.", &spb);
            res = false;
            break;
        }
        break;
    case TI_FLT32:
        switch (value_type.element_type) {
        case ET_VALUE:
            glUniform1fv(loc, GLsizei(nr_elements), reinterpret_cast<const flt32_type*>(value_ptr));
            break;
        case ET_VECTOR:
            switch (value_type.nr_rows) {
            case 2: glUniform2fv(loc, GLsizei(nr_elements), reinterpret_cast<const flt32_type*>(value_ptr)); break;
            case 3: glUniform3fv(loc, GLsizei(nr_elements), reinterpret_cast<const flt32_type*>(value_ptr)); break;
            case 4: glUniform4fv(loc, GLsizei(nr_elements), reinterpret_cast<const flt32_type*>(value_ptr)); break;
            default:
                error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") vector dimension outside [2,..4].", &spb);
                res = false;
                break;
            }
            break;
        case ET_MATRIX:
            switch (value_type.nr_rows) {
            case 2:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix2fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 3: glUniformMatrix2x3fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 4: glUniformMatrix2x4fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..4].", &spb);
                    res = false;
                    break;
                }
                break;
            case 3:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix3x2fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 3: glUniformMatrix3fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 4: glUniformMatrix3x4fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                default:
                    error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..4].", &spb);
                    res = false;
                    break;
                }
                break;
            case 4:
                switch (value_type.nr_columns) {
                case 2: glUniformMatrix4x2fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 3: glUniformMatrix4x3fv(loc, GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr)); break;
                case 4: glUniformMatrix4fv(loc,   GLsizei(nr_elements), value_type.is_row_major, reinterpret_cast<const flt32_type*>(value_ptr));   break;
                default:
                    error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") matrix number of columns outside [2,..4].", &spb);
                    res = false;
                    break;
                }
                break;
            default:
                error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") matrix number of rows outside [2,..4].", &spb);
                res = false;
                break;
            }
            break;
        }
        break;
    default:
        error(std::string("gl_context::set_uniform_array_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type (only int32, uint32, and flt32 supported).", &spb);
        res = false;
        break;
    }
    if (check_gl_error("gl_context::set_uniform_array_void()", &spb))
        res = false;
 
    if (not_current)
        glUseProgram(shader_program_stack.empty() ? 0 : get_gl_id(shader_program_stack.top()->handle));
 
    return res;
}
 
int  gl_context::get_uniform_block_index(const shader_program_base& spb, const std::string& name) const
{
    return glGetUniformBlockIndex(get_gl_id(spb.handle), name.c_str());
}
 
bool gl_context::set_uniform_block_binding(shader_program_base& spb, int index, int binding) const
{
    if(!spb.handle) {
        error("gl_context::set_uniform_block_binding() called on not created program", &spb);
        return false;
    }
    glUniformBlockBinding(get_gl_id(spb.handle), index, binding);
    return check_gl_error("gl_context::set_uniform_block_binding()", &spb);
}
 
int  gl_context::get_attribute_location(const shader_program_base& spb, const std::string& name) const
{
    return glGetAttribLocation(get_gl_id(spb.handle), name.c_str());
}
 
bool gl_context::set_attribute_void(shader_program_base& spb, int loc, type_descriptor value_type, const void* value_ptr) const
{
    if (!spb.handle) {
        error("gl_context::set_attribute_void() called on not created program", &spb);
        return false;
    }
    bool not_current = shader_program_stack.empty() || shader_program_stack.top() != &spb;
    if (not_current)
        glUseProgram(get_gl_id(spb.handle));
    bool res = true;
    switch (value_type.element_type) {
    case ET_VALUE:
        switch (value_type.coordinate_type) {
        case TI_BOOL:   glVertexAttrib1s(loc, *reinterpret_cast<const bool*>(value_ptr) ? 1 : 0); break;
        case TI_INT8:   glVertexAttrib1s(loc, *reinterpret_cast<const int8_type*>(value_ptr)); break;
        case TI_INT16:  glVertexAttrib1s(loc, *reinterpret_cast<const int16_type*>(value_ptr)); break;
        case TI_INT32:  glVertexAttribI1i(loc, *reinterpret_cast<const int32_type*>(value_ptr)); break;
        case TI_UINT8:  glVertexAttrib1s(loc, *reinterpret_cast<const uint8_type*>(value_ptr)); break;
        case TI_UINT16: glVertexAttribI1ui(loc, *reinterpret_cast<const uint16_type*>(value_ptr)); break;
        case TI_UINT32: glVertexAttribI1ui(loc, *reinterpret_cast<const uint32_type*>(value_ptr)); break;
        case TI_FLT32:  glVertexAttrib1f(loc, *reinterpret_cast<const flt32_type*>(value_ptr)); break;
        case TI_FLT64:  glVertexAttrib1d(loc, *reinterpret_cast<const flt64_type*>(value_ptr)); break;
        default:
            error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") type not supported!", &spb);
            res = false;
            break;
        }
        break;
    case ET_VECTOR:
        switch (value_type.nr_rows) {
        case 2:
            switch (value_type.coordinate_type) {
            case TI_BOOL:    glVertexAttrib2s(loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0); break;
            case TI_UINT8:   glVertexAttrib2s(loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1]); break;
            case TI_UINT16: glVertexAttribI2ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1]); break;
            case TI_UINT32: glVertexAttribI2ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1]); break;
            case TI_INT8:    glVertexAttrib2s(loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1]); break;
            case TI_INT16:   glVertexAttrib2s(loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1]); break;
            case TI_INT32:  glVertexAttribI2i(loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1]); break;
            case TI_FLT32:   glVertexAttrib2f(loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1]); break;
            case TI_FLT64:   glVertexAttrib2d(loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1]); break;
            default:
                error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false;
                break;
            }
            break;
        case 3:
            switch (value_type.coordinate_type) {
            case TI_BOOL:    glVertexAttrib3s (loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[2] ? 1 : 0); break;
            case TI_UINT8:   glVertexAttrib3s (loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1], reinterpret_cast<const uint8_type*> (value_ptr)[2]); break;
            case TI_UINT16: glVertexAttribI3ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1], reinterpret_cast<const uint16_type*>(value_ptr)[2]); break;
            case TI_UINT32: glVertexAttribI3ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1], reinterpret_cast<const uint32_type*>(value_ptr)[2]); break;
            case TI_INT8:    glVertexAttrib3s (loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1], reinterpret_cast<const int8_type*>  (value_ptr)[2]); break;
            case TI_INT16:   glVertexAttrib3s (loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1], reinterpret_cast<const int16_type*> (value_ptr)[2]); break;
            case TI_INT32:  glVertexAttribI3i (loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1], reinterpret_cast<const int32_type*> (value_ptr)[2]); break;
            case TI_FLT32:   glVertexAttrib3f (loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1], reinterpret_cast<const flt32_type*> (value_ptr)[2]); break;
            case TI_FLT64:   glVertexAttrib3d (loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1], reinterpret_cast<const flt64_type*> (value_ptr)[2]); break;
            default:
                error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false; break;
            }
            break;
        case 4:
            switch (value_type.coordinate_type) {
            case TI_BOOL:    glVertexAttrib4s (loc, reinterpret_cast<const bool*>(value_ptr)[0] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[1] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[2] ? 1 : 0, reinterpret_cast<const bool*>(value_ptr)[3] ? 1 : 0); break;
            case TI_UINT8:   glVertexAttrib4s (loc, reinterpret_cast<const uint8_type*> (value_ptr)[0], reinterpret_cast<const uint8_type*> (value_ptr)[1], reinterpret_cast<const uint8_type*> (value_ptr)[2], reinterpret_cast<const uint8_type*> (value_ptr)[3]); break;
            case TI_UINT16: glVertexAttribI4ui(loc, reinterpret_cast<const uint16_type*>(value_ptr)[0], reinterpret_cast<const uint16_type*>(value_ptr)[1], reinterpret_cast<const uint16_type*>(value_ptr)[2], reinterpret_cast<const uint16_type*>(value_ptr)[3]); break;
            case TI_UINT32: glVertexAttribI4ui(loc, reinterpret_cast<const uint32_type*>(value_ptr)[0], reinterpret_cast<const uint32_type*>(value_ptr)[1], reinterpret_cast<const uint32_type*>(value_ptr)[2], reinterpret_cast<const uint32_type*>(value_ptr)[3]); break;
            case TI_INT8:    glVertexAttrib4s (loc, reinterpret_cast<const int8_type*>  (value_ptr)[0], reinterpret_cast<const int8_type*>  (value_ptr)[1], reinterpret_cast<const int8_type*>  (value_ptr)[2], reinterpret_cast<const int8_type*>  (value_ptr)[3]); break;
            case TI_INT16:   glVertexAttrib4s (loc, reinterpret_cast<const int16_type*> (value_ptr)[0], reinterpret_cast<const int16_type*> (value_ptr)[1], reinterpret_cast<const int16_type*> (value_ptr)[2], reinterpret_cast<const int16_type*> (value_ptr)[3]); break;
            case TI_INT32:  glVertexAttribI4i (loc, reinterpret_cast<const int32_type*> (value_ptr)[0], reinterpret_cast<const int32_type*> (value_ptr)[1], reinterpret_cast<const int32_type*> (value_ptr)[2], reinterpret_cast<const int32_type*> (value_ptr)[3]); break;
            case TI_FLT32:   glVertexAttrib4f (loc, reinterpret_cast<const flt32_type*> (value_ptr)[0], reinterpret_cast<const flt32_type*> (value_ptr)[1], reinterpret_cast<const flt32_type*> (value_ptr)[2], reinterpret_cast<const flt32_type*> (value_ptr)[3]); break;
            case TI_FLT64:   glVertexAttrib4d (loc, reinterpret_cast<const flt64_type*> (value_ptr)[0], reinterpret_cast<const flt64_type*> (value_ptr)[1], reinterpret_cast<const flt64_type*> (value_ptr)[2], reinterpret_cast<const flt64_type*> (value_ptr)[3]); break;
            default:
                error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") unsupported coordinate type.", &spb);
                res = false;
                break;
            }
            break;
        default:
            error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") vector dimension outside [2..4]", &spb);
            res = false;
            break;
        }
        break;
    case ET_MATRIX:
        error(std::string("gl_context::set_attribute_void(") + value_type_index_to_string(value_type) + ") matrix type not supported!", &spb);
        res = false;
        break;
    }
    if (not_current)
        glUseProgram(shader_program_stack.empty() ? 0 : get_gl_id(shader_program_stack.top()->handle));
 
    if (check_gl_error("gl_context::set_uniform_array_void()", &spb))
        res = false;
    return res;
}
 
bool gl_context::attribute_array_binding_create(attribute_array_binding_base& aab) const
{
    if (!GLEW_VERSION_3_0) {
        error("gl_context::attribute_array_binding_create() array attribute bindings not supported", &aab);
        return false;
    }
    GLuint a_id;
    glGenVertexArrays(1, &a_id);
    if (a_id == -1) {
        error(std::string("gl_context::attribute_array_binding_create(): ") + gl_error(), &aab);
        return false;
    }
    aab.ctx_ptr = this;
    aab.handle = get_handle(a_id);
    return true;
}
 
bool gl_context::attribute_array_binding_destruct(attribute_array_binding_base& aab) const
{
    if (!attribute_array_binding_stack.empty() && &aab == attribute_array_binding_stack.top())
        glBindVertexArray(0);
    if (!context::attribute_array_binding_destruct(aab))
        return false;
    if (!aab.handle) {
        error("gl_context::attribute_array_binding_destruct(): called on not created attribute array binding", &aab);
        return false;
    }
    GLuint a_id = get_gl_id(aab.handle);
    glDeleteVertexArrays(1, &a_id);
    aab.handle = 0;
    aab.ctx_ptr = 0;
    return !check_gl_error("gl_context::attribute_array_binding_destruct");
}
 
bool gl_context::attribute_array_binding_enable(attribute_array_binding_base& aab)
{
    if (!context::attribute_array_binding_enable(aab))
        return false;
    glBindVertexArray(get_gl_id(aab.handle));
    return !check_gl_error("gl_context::attribute_array_binding_enable");
}
 
bool gl_context::attribute_array_binding_disable(attribute_array_binding_base& aab)
{
    if (!context::attribute_array_binding_disable(aab))
        return false;
    if (attribute_array_binding_stack.empty())
        glBindVertexArray(0);
    else
        glBindVertexArray(get_gl_id(attribute_array_binding_stack.top()->handle));
    return true;
}
 
bool gl_context::set_element_array(attribute_array_binding_base* aab, const vertex_buffer_base* vbb) const
{
    if (!vbb) {
        error("gl_context::set_element_array(): called without a vertex buffer object", aab);
        return false;
    }
    if (!vbb->handle) {
        error("gl_context::set_element_array(): called with not created vertex buffer object", vbb);
        return false;
    }
    if (vbb->type != VBT_INDICES) {
        std::cout << "gl_context::set_element_array() : called on vertex buffer object that is not of type VBT_INDICES" << std::endl;
//      error("gl_context::set_element_array(): called on vertex buffer object that is not of type VBT_INDICES", vbb);
//      return false;
    }
    if (aab) {
        if (!aab->handle) {
            error("gl_context::set_element_array(): called on not created attribute array binding", aab);
            return false;
        }
    }
    // enable vertex array
    bool not_current = attribute_array_binding_stack.empty() || attribute_array_binding_stack.top() != aab;
    if (aab && not_current)
        glBindVertexArray(get_gl_id(aab->handle));
 
    // bind buffer to element array
    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, get_gl_id(vbb->handle));
 
    if (aab && not_current)
        glBindVertexArray(attribute_array_binding_stack.empty() ? 0 : get_gl_id(attribute_array_binding_stack.top()->handle));
 
    return !check_gl_error("gl_context::set_element_array_void()", aab);
}
 
 
bool gl_context::set_attribute_array_void(attribute_array_binding_base* aab, int loc, type_descriptor value_type, const vertex_buffer_base* vbb, const void* ptr, size_t nr_elements, unsigned stride) const
{
    if (value_type == ET_MATRIX) {
        error("gl_context::set_attribute_array_void(): called with matrix elements not supported", aab);
        return false;
    }
    if (vbb) {
        if (!vbb->handle) {
            error("gl_context::set_attribute_array_void(): called with not created vertex buffer object", vbb);
            return false;
        }
    }
    if (aab) {
        if (!aab->handle) {
            error("gl_context::set_attribute_array_void(): called on not created attribute array binding", aab);
            return false;
        }
    }
 
    bool not_current = attribute_array_binding_stack.empty() || attribute_array_binding_stack.top() != aab;
    if (aab && not_current)
        glBindVertexArray(get_gl_id(aab->handle));
 
    if (vbb)
        glBindBuffer(GL_ARRAY_BUFFER, get_gl_id(vbb->handle));
 
    bool res = true;
    unsigned n = value_type.element_type == ET_VALUE ? 1 : value_type.nr_rows;
    switch (value_type.coordinate_type) {
    case TI_INT8: value_type.normalize ? glVertexAttribPointer(loc, n, GL_BYTE, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_BYTE, stride, ptr); break;
    case TI_INT16:  value_type.normalize ? glVertexAttribPointer(loc, n, GL_SHORT, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_SHORT, stride, ptr); break;
    case TI_INT32:  value_type.normalize ? glVertexAttribPointer(loc, n, GL_INT, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_INT, stride, ptr); break;
    case TI_UINT8:  value_type.normalize ? glVertexAttribPointer(loc, n, GL_UNSIGNED_BYTE, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_UNSIGNED_BYTE, stride, ptr); break;
    case TI_UINT16: value_type.normalize ? glVertexAttribPointer(loc, n, GL_UNSIGNED_SHORT, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_UNSIGNED_SHORT, stride, ptr); break;
    case TI_UINT32: value_type.normalize ? glVertexAttribPointer(loc, n, GL_UNSIGNED_INT, value_type.normalize, stride, ptr) : glVertexAttribIPointer(loc, n, GL_UNSIGNED_INT, stride, ptr); break;
    case TI_FLT32: glVertexAttribPointer(loc, n, GL_FLOAT, value_type.normalize, stride, ptr); break;
    case TI_FLT64:
        if (GLEW_VERSION_4_1)
            glVertexAttribLPointer(loc, n, GL_DOUBLE, stride, ptr);
        else {
            error("gl_context::set_attribute_array_void(): called with coordinates of type double only supported starting with OpenGL 4.1", aab);
            res = false;
        }
        break;
    default:
        error("gl_context::set_attribute_array_void(): called with unsupported coordinate type", aab);
        res = false;
    }
 
    if (res)
        glEnableVertexAttribArray(loc);
 
    if (vbb)
        glBindBuffer(GL_ARRAY_BUFFER, 0);
 
    if (aab && not_current)
        glBindVertexArray(attribute_array_binding_stack.empty() ? 0 : get_gl_id(attribute_array_binding_stack.top()->handle));
 
 
    return res && !check_gl_error("gl_context::set_attribute_array_void()", aab);
}
 
bool gl_context::enable_attribute_array(attribute_array_binding_base* aab, int loc, bool do_enable) const
{
    bool not_current = attribute_array_binding_stack.empty() || attribute_array_binding_stack.top() != aab;
    if (aab) {
        if (!aab->handle) {
            error("gl_context::enable_attribute_array(): called on not created attribute array binding", aab);
            return false;
        }
        if (not_current)
            glBindVertexArray(get_gl_id(aab->handle));
    }
 
    if (do_enable)
        glEnableVertexAttribArray(loc);
    else
        glDisableVertexAttribArray(loc);
 
    if (aab && not_current)
        glBindVertexArray(attribute_array_binding_stack.empty() ? 0 : get_gl_id(attribute_array_binding_stack.top()->handle));
 
    return !check_gl_error("gl_context::enable_attribute_array()");
}
 
bool gl_context::is_attribute_array_enabled(const attribute_array_binding_base* aab, int loc) const
{
    bool not_current = attribute_array_binding_stack.empty() || attribute_array_binding_stack.top() != aab;
    if (aab) {
        if (!aab->handle) {
            error("gl_context::is_attribute_array_enabled(): called on not created attribute array binding", aab);
            return false;
        }
        if (not_current)
            glBindVertexArray(get_gl_id(aab->handle));
    }
 
    GLint res;
    glGetVertexAttribiv(loc, GL_VERTEX_ATTRIB_ARRAY_ENABLED, &res);
 
    if (aab && not_current)
        glBindVertexArray(attribute_array_binding_stack.empty() ? 0 : get_gl_id(attribute_array_binding_stack.top()->handle));
 
    return res == GL_TRUE;
}
 
GLenum buffer_target(VertexBufferType vbt)
{
    static GLenum buffer_targets[] = {
        GL_ARRAY_BUFFER,
        GL_ELEMENT_ARRAY_BUFFER,
        GL_TEXTURE_BUFFER,
        GL_UNIFORM_BUFFER,
        GL_TRANSFORM_FEEDBACK_BUFFER,
        GL_SHADER_STORAGE_BUFFER,
        GL_ATOMIC_COUNTER_BUFFER,
        GL_DRAW_INDIRECT_BUFFER
    };
    return buffer_targets[vbt];
}
 
GLenum buffer_usage(VertexBufferUsage vbu)
{
    static GLenum buffer_usages[] = {
        GL_STREAM_DRAW,
        GL_STREAM_READ,
        GL_STREAM_COPY,
        GL_STATIC_DRAW,
        GL_STATIC_READ,
        GL_STATIC_COPY,
        GL_DYNAMIC_DRAW,
        GL_DYNAMIC_READ,
        GL_DYNAMIC_COPY
    };
    return buffer_usages[vbu];
}
 
bool gl_context::vertex_buffer_bind(const vertex_buffer_base& vbb, VertexBufferType _type, unsigned _idx) const
{
    if (_idx == unsigned(-1))
        glBindBuffer(buffer_target(_type), get_gl_id(vbb.handle));
    else
        glBindBufferBase(buffer_target(_type), _idx, get_gl_id(vbb.handle));
    return !check_gl_error("gl_context::vertex_buffer_bind", &vbb);
}
 
bool gl_context::vertex_buffer_unbind(const vertex_buffer_base& vbb, VertexBufferType _type, unsigned _idx) const {
    if(_idx == unsigned(-1))
        glBindBuffer(buffer_target(_type), 0);
    else
        glBindBufferBase(buffer_target(_type), _idx, 0);
    return !check_gl_error("gl_context::vertex_buffer_unbind", &vbb);
}
 
bool gl_context::vertex_buffer_create(vertex_buffer_base& vbb, const void* array_ptr, size_t size_in_bytes) const
{
    if (!GLEW_VERSION_2_0) {
        error("gl_context::vertex_buffer_create() vertex buffer objects not supported", &vbb);
        return false;
    }
    GLuint b_id;
    glGenBuffers(1, &b_id);
    if (b_id == -1) {
        error(std::string("gl_context::vertex_buffer_create(): ") + gl_error(), &vbb);
        return false;
    }
    vbb.handle = get_handle(b_id);
    glBindBuffer(buffer_target(vbb.type), b_id);
    glBufferData(buffer_target(vbb.type), size_in_bytes, array_ptr, buffer_usage(vbb.usage));
    glBindBuffer(buffer_target(vbb.type), 0);
    return !check_gl_error("gl_context::vertex_buffer_create", &vbb);
}
 
bool gl_context::vertex_buffer_resize(vertex_buffer_base& vbb, const void* array_ptr, size_t size_in_bytes) const {
    if(!vbb.handle) {
        error("gl_context::vertex_buffer_resize() vertex buffer object must be created before", &vbb);
        return false;
    }
    GLuint b_id = get_gl_id(vbb.handle);
    glBindBuffer(buffer_target(vbb.type), b_id);
    glBufferData(buffer_target(vbb.type), size_in_bytes, array_ptr, buffer_usage(vbb.usage));
    glBindBuffer(buffer_target(vbb.type), 0);
    return !check_gl_error("gl_context::vertex_buffer_resize", &vbb);
}
 
bool gl_context::vertex_buffer_clear(vertex_buffer_base& vbb, size_t offset, size_t size_in_bytes) const
{
    if (!vbb.handle) {
        error("gl_context::vertex_buffer_clear() vertex buffer object must be created before", &vbb);
        return false;
    }
    GLuint b_id = get_gl_id(vbb.handle);
    glBindBuffer(buffer_target(vbb.type), b_id);
    glClearBufferSubData(buffer_target(vbb.type), GL_R8, offset, size_in_bytes, GL_RED, GL_UNSIGNED_BYTE, NULL);
    glBindBuffer(buffer_target(vbb.type), 0);
    return !check_gl_error("gl_context::vertex_buffer_clear", &vbb);
}
 
bool gl_context::vertex_buffer_replace(vertex_buffer_base& vbb, size_t offset, size_t size_in_bytes, const void* array_ptr) const
{
    if (!vbb.handle) {
        error("gl_context::vertex_buffer_replace() vertex buffer object must be created before", &vbb);
        return false;
    }
    GLuint b_id = get_gl_id(vbb.handle);
    glBindBuffer(buffer_target(vbb.type), b_id);
    glBufferSubData(buffer_target(vbb.type), offset, size_in_bytes, array_ptr);
    glBindBuffer(buffer_target(vbb.type), 0);
    return !check_gl_error("gl_context::vertex_buffer_replace", &vbb);
}
 
bool gl_context::vertex_buffer_copy(const vertex_buffer_base& src, size_t src_offset, vertex_buffer_base& target, size_t target_offset, size_t size_in_bytes) const
{
    if (!src.handle || !target.handle) {
        error("gl_context::vertex_buffer_copy() source and destination vertex buffer objects must have been created before", &src);
        return false;
    }
    GLuint b_id = get_gl_id(src.handle);
    glBindBuffer(GL_COPY_READ_BUFFER, get_gl_id(src.handle));
    glBindBuffer(GL_COPY_WRITE_BUFFER, get_gl_id(target.handle));
    glCopyBufferSubData(GL_COPY_READ_BUFFER, GL_COPY_WRITE_BUFFER, src_offset, target_offset, size_in_bytes);
    glBindBuffer(GL_COPY_READ_BUFFER, 0);
    glBindBuffer(GL_COPY_WRITE_BUFFER, 0);
    return !check_gl_error("gl_context::vertex_buffer_copy", &src);
 
}
 
bool gl_context::vertex_buffer_copy_back(const vertex_buffer_base& vbb, size_t offset, size_t size_in_bytes, void* array_ptr) const
{
    if (!vbb.handle) {
        error("gl_context::vertex_buffer_copy_back() vertex buffer object must be created", &vbb);
        return false;
    }
    GLuint b_id = get_gl_id(vbb.handle);
    GLuint bind_point = GL_COPY_READ_BUFFER;
    switch (vbb.type) {
    case cgv::render::VBT_VERTICES: bind_point = GL_ARRAY_BUFFER; break;
    case cgv::render::VBT_INDICES: bind_point = GL_ELEMENT_ARRAY_BUFFER; break;
    case cgv::render::VBT_TEXTURE: bind_point = GL_TEXTURE_BUFFER; break;
    case cgv::render::VBT_UNIFORM: bind_point = GL_UNIFORM_BUFFER; break;
    case cgv::render::VBT_FEEDBACK: bind_point = GL_TRANSFORM_FEEDBACK_BUFFER; break;
    case cgv::render::VBT_STORAGE: bind_point = GL_SHADER_STORAGE_BUFFER; break;
    case cgv::render::VBT_ATOMIC_COUNTER: bind_point = GL_ATOMIC_COUNTER_BUFFER; break;
    case cgv::render::VBT_INDIRECT: bind_point = GL_DRAW_INDIRECT_BUFFER; break;
    }
    glBindBuffer(bind_point, b_id);
    glGetBufferSubData(bind_point, offset, size_in_bytes, array_ptr);
    glBindBuffer(bind_point, 0);
    return !check_gl_error("gl_context::vertex_buffer_copy_back", &vbb);
}
 
bool gl_context::vertex_buffer_destruct(vertex_buffer_base& vbb) const
{
    if (vbb.handle) {
        GLuint b_id = get_gl_id(vbb.handle);
        glDeleteBuffers(1, &b_id);
        vbb.handle = 0;
        vbb.ctx_ptr = 0;
        return !check_gl_error("gl_context::vertex_buffer_destruct");
    }
    else {
        error("gl_context::vertex_buffer_destruct(): called on not created vertex buffer", &vbb);
        return false;
    }
}
 
 
        }
    }
}