This is a somewhat large update to the inner working of GlslProg and a smaller update to the inner workings of Batch and VboMesh. Much of the public facing interface hasn't changed at all.

The highlights in GlslProg are:

• Clean up of caching variables. We now store Uniforms as a unified struct, same with attributes, uniform blocks, and transform feedback varyings.
• Uniforms now have a CPU-side cache so as to not re-buffer uniform values that haven't changed.
• We don't lazily initialize cache variables anymore. All of the above are cached during initialization so as to mitigate OpenGL get* operations.
• Using templated type checking for uniform functions.
• Added multiple uniform type overloads.
• Added public facing getters for all of the new structs of info. Added public facing const pointer find methods for the new structs.

The highlights in Batch and VboMesh are:

• Batch - removed initVao logic to centralize Buffer/GlslProg Vao creation inside vboMesh, in buildVao function.
• attribs that have a size greater than 4 now initialize a vao correctly. For instance, sending a mat4 up to the gpu.

Also changed:

• Cloth Simulation sample updated. Found a bug with AMD GPUs that wasn't allowing two TFO's, which is in progress. However, I made the sample work on all platforms including AMD until by using gl::bindBufferBase until AMD fixes the problem.
• Changed ConstantStrings.h/.cpp to ConstantConversions.h/.cpp to include other conversions such as size of conversions.

I've got this working in Xcode now (by downgrading to GLSL 410). I also got rid of the lambdas in ::draw() that were no longer necessary and replaced window setup with scoped objects.

This PR adds optional support for the GStreamer player under MSW 64bit through the CMake build system backend. It has only been tested on a VM running Windows 10 with Visual Studio 2015 so I m putting it here in case someone would like to give it a proper try.

You will need to have GStreamer installed and available on the environment in order for this to work. The latest version as of now is 1.12.4. You will need gstreamer-1.0-devel-x86_64-1.12.4.msi and gstreamer-1.0-x86_64-1.12.4.msi .

Choose the complete installation when prompt in order to get all necessary GStreamer plugins installed.

Once the installation is complete you should add the GStreamer installation bin and lib paths to your PATH env variable. By default these should be C:\gstreamer\1.0\x86_64\bin and C:\gstreamer\1.0\x86_64\lib

To enable it you will need to generate the VS files for Cinder with the following:

cmake .. -G "Visual Studio 14 2015 Win64" -DCINDER_MSW_USE_GSTREAMER=1

For testing you can build the QuickTimeBasic sample by generating VS project files through:

cmake .. -G "Visual Studio 14 2015 Win64"

Rebased the various docs commits mentioned in https://github.com/cinder/Cinder/issues/1213 onto the release_v0.9.0 branch so we can update the website.

Fixes #1135 & #1245.

Here's the git rebase -i origin/release_v0.9.0 commands I used:

#1148 Updated git guide. Fixed outdated/broken links.

pick f0c04ba49cfbe91a224132e4af184ca4d552dab3 Updated index docs links

pick f1b930f432582c01846651d379251ec3de15a8c6 Merge to resolve conflicts

#1149 Added logging guide

pick 6d1220fdb9e5281c174557a2d5469901900eccbf Added logging guide pick 5c3eb4212a2b6147b06e52dd1a07e586cb504b6e Formatting updates based on feedback #1150 Added supplementary html for LoggerFile and LoggerFileRotating

pick 0bab421cbc4484dbcd9d8639a3a9380a2f507701 Added supplementary html for LoggerFile and LoggerFileRotating pick 8968427740f131d445a7742a2b2b8366f4f422b4 small formatting changes. #1155 Dox next fixes

pick 2820edbb23a60e95fceeb684a59dd73c18474928 No more overriding key events after click occurs.

pick 1f0eeead8ef15dca7784e27ad523dfbe974588d2 Merge to resolve conflicts

pick 71feb3ed6819c114f81c56a5fb9d3590f2aa6b1c Empty without merge resolution

#1177 Fixed the keyboard input bug for all pages

pick c388e3233acb567f469a0c0a922beb5cf45df369 Removed input focus on key input. #1183 Fix download links and migrate links to https

pick a1b89a9f0517057952c4e03b3a06eb677f83a60e fix download links and migrate links to https #1155 Dox next fixes

pick a75048b Tweaked label for 0.9.1.dev link #1197 Update TinderBox link

pick f1a0fdf65d35bff0e268ae675f703f9b38a8b3f3 Update TinderBox link #1203 Fixed code syntax within function descriptions

pick 6be9188b1aaf88a20eb1c726d3e368b440ffdd72 Fixed code syntax within funciton descriptions. #1216 Added logging guide to main guide page.

pick d49732f67f5fbb5b6c3f3220a6c31d0ca57af0d9 Added logging guide to main guide page. pick c4e55c8f023e7781a019c48e616f93c4363b31f4 Spaces to tabs for consistency #1233 Audio doc cleanup

pick e8e9e50edef1b3c49dc36460032bb547de4a6996 remove stray code section in audio guide pick 9bc804418cd337a25adc231cb43d88ae643065ac fix doc for updated function name

There were a couple small tweaks I had to make to resolve conflicts... I think it's what the battling merge commits were trying to do.

This PR will replace glLoad with the GLAD extension loader, allowing us to add OpenGL v4.6 features to Cinder. glLoad has not been updated in a while, so we're looking for an alternative.

When the GLAD files are generated without additional extensions, Cinder users will be forced to create a RendererGl of the correct version like so: CINDER_APP( MyApp, RendererGl( RendererGl::Options().version( 4, 6 ) ) ) to use the desired features.

To prevent confusion, however, we should add commonly used extensions when generating the GLAD files. I've added a minimal set of extensions, enough to make Cinder work out of the box. Before accepting this PR, it would be smart to discuss what extra extensions we might need.

GLAD features a webservice that makes it extremely easy to update the source files. See glad/glad.h, which is the main header of the loader. At the top, you'll find the exact settings used to create the files. If we want to add additional extensions, use the provided information as a starting point for the webservice and generate a new header.

An alternative to GLAD is FlextGL, which requires Python to generate the files. It would allow us to automate the process, if we're Python savvy, that is. But I'd say that GLAD's webservice is a much easier alternative which does not require us to install any additional development software.

The Linux version has not been tested and probably needs a few fixes to work. However, I have included the required GLX files, so this should not be too hard.

When adding settings->setFullScreen( true ); to prepareSettings() in OS X 10.10, the following warning / stacktrace is printed (tested on master branch):

2014-11-14 19:32:52.227 windowTest[85547:5829855] NSWindow warning: adding an unknown subview: <NSView: 0x10645c0>
2014-11-14 19:32:52.244 windowTest[85547:5829855] Call stack:
(
    0   AppKit                              0x917fa776 -[NSThemeFrame addSubview:] + 108
    1   AppKit                              0x910a1d77 -[NSView replaceSubview:with:] + 152
    2   AppKit                              0x9192b4d7 __60-[NSView(NSFullScreenMode) enterFullScreenMode:withOptions:]_block_invoke + 178
    3   AppKit                              0x91191bd7 NSPerformWithScreenUpdatesDisabled + 97
    4   AppKit                              0x9192b3cd -[NSView(NSFullScreenMode) enterFullScreenMode:withOptions:] + 3147
    5   windowTest                          0x0007d00c -[CinderView setFullScreen:options:] + 1036
    6   windowTest                          0x000430e7 -[WindowImplBasicCocoa setFullScreen:options:] + 135
    7   windowTest                          0x00040d1b -[AppImplCocoaBasic applicationDidFinishLaunching:] + 1403

Though the message is massively ugly, it does appear to still work. @andrewfb tried using the NSFullScreenModeSetting setting, which he found in Apple's OpenCL N-Body sample, but this seems to stop full screen from working altogether.

I moved the advanced deferred sample into a nifty pentagonal room using geom::Cylinder. I'm getting some weird banding from it and a weird stripe sometimes. In the screenshots, you'll see that the floor and ceiling reflect OK, but the walls do not. Check out the debug view where it's just the normals. You can see a weird strip occuring. I'm using the same shader to render everything to the G-buffer, so it's definitely something weird with just this one shape.

There are a few updates I'd like to make to the cmake files, in order to make it easier for the user (and or a package management system) to control the cinder build via typical cmake configuration options. Two examples being the control of the used c++ standard for which CMAKE provides the CMAKE_CXX_STANDARD variable since cmake 3.1 or the hardcoded usage of MT (https://github.com/cinder/Cinder/issues/2116)

I think that would also be a good time to upgrade the minimal required cmake version, as 2.8 is really old and a lot of things have become easier since then.

I wanted to ask, what an acceptable baseline would be. To give a short overview: VS2019 currently ships with 3.15, Ubuntu 18.04 with cmake 3.10, Ubuntu16.04 with 3.5. Travis uses 3.12. I have no idea about macos, but generally it is very easy to install a newer version of cmake on any system.

Depending on whether Ubuntu 16.04 is still a common target, I'd recommend 3.5 or 3.10 (among other things, cmake 3.8 adds the cxx_std_14 / 17 target properties which are "nice to have" but not critical).

I wanted to extend the build matrix for Linux on Travis CI, and was wondering what toolchains / to versions cinder aims to support.

For example, the cmake files claim to require only cmake 2.8, Ubuntu 14.04 however comes with 2.8.12, so 2.8 seems overly restrictive. The Travis-CI images come with preinstalled 3.9, and the dependency script installs 3.2. The situation with the compilers is similarly unclear. Is gcc supposed to be supported at all? Which version?

Updating a block sample to glNext latest, I faced some uniform problems. My mat4 array uniform would not be recognized, and the data would never reach the gpu. AFAIK, the problem is that glGetActiveUniform appends '[0]' to the array uniform names, which are stored as is. Later when findUniform is called, no match is found.

Previously, findUniform would call glGetUniformLocation directly, which would still work because locations "can be retrieved by using the name of the array, or by using the name appended by "[0]". http://docs.gl/gl3/glGetUniformLocation

I did temporarily fix this problem by storing the name without '[0]':

inline bool ends_with( std::string const & value, std::string const & ending )
{
    if( ending.size() > value.size() ) return false;
    return std::equal( ending.rbegin(), ending.rend(), value.rbegin() );
}

Uniform uniform;
uniform.mName           = name;
if( ends_with( name, "[0]" ) ) {
    uniform.mName = uniform.mName.substr( 0, uniform.mName.length() - 3 );
}

I'm not sure yet what the optimal solution would be.

Note that the new mUniforms is modified in several places: https://github.com/cinder/Cinder/blob/glNext/src/cinder/gl/GlslProg.cpp#L735 https://github.com/cinder/Cinder/blob/glNext/src/cinder/gl/GlslProg.cpp#L275 https://github.com/cinder/Cinder/blob/glNext/src/cinder/gl/GlslProg.cpp#L343 https://github.com/cinder/Cinder/blob/glNext/src/cinder/gl/GlslProg.cpp#L364

Opening last PR to finally get Emscripten working!

Should be considered alpha state - still needs to be tested a little more since things were hand-imported for simplicity. If you'd like to help test you can of course pull this PR or my branch of Cinder. Generate and read the docs for the full details, but in short, with the exception of sample building, you can follow ChaoticBob's old instructions and from there, it's like a normal CMake based desktop project. See <repo>/samples/_emscripten for sample CMake files you can use.

Note that audio stuff is still somewhat in flux.

Also should note that most of the core emscripten stuff is header only - will move some stuff to implementation files at a later point.

Any changes / fixes going forward will go to this PR

Description: After the MSVC team implemented P2468R2 The Equality Operator You Are Looking For, we got a compiler error with /std:c++latest like below, the error is due to the equality operator here https://github.com/cinder/Cinder/blob/master/src/AntTweakBar/TwMgr.h#L369 needs a 'const'. Could you please look this issue? Thanks.

F:\gitP\cinder\Cinder\src\AntTweakBar\TwMgr.cpp(2176,71): error C2666: 'CRect::operator ==': overloaded functions have similar conversions [F:\gitP\cinder\Cinder\build_amd64\cinder.vcxproj]

Note: this issue will be reproduced on next release version of VS(VS17.6 or later)

Repro steps:

1. open VC2019 x64 tools command
2. git clone https://github.com/cinder/Cinder F:\gitP\cinder\Cinder
3. mkdir F:\gitP\cinder\Cinder\build_amd64
4. cd F:\gitP\cinder\Cinder\build_amd64
5. cmake -G "Visual Studio 16 2019" -A x64 -DCMAKE_SYSTEM_VERSION=10.0.18362.0 -DCMAKE_BUILD_TYPE=Release -DCINDER_BUILD_TESTS=ON ..
6. set CL= /std:c++latest /D_HAS_DEPRECATED_UNCAUGHT_EXCEPTION=1 /D_SILENCE_CXX17_UNCAUGHT_EXCEPTION_DEPRECATION_WARNING
7. msbuild /m /p:Platform=x64 /p:Configuration=Release cinder.sln /t:Rebuild > build.log

Detailed log: build.log

EqualizerNode is a FilterNode which supports an arbitrary number of filters, while the FilterBiquadNode only supports a single filter.

This also fixes quite a few bugs in shelving and low/high pass filters... Please tell me what you think

When trying to find out which modifier keys were pressed during a fileDrop() event, I found it was necessary to "roll my own" custom solution. (see https://discourse.libcinder.org/t/getting-keyboard-modifiers-in-filedrop/2043)

It would be nice if this cross-platform functionality was exposed as a Cinder function (as @axjxwright wrote, this can be done using Cocoa: [NSEvent modifierFlags] and Windows: GetAsyncKeyState()).

This pull request implements a collection of fixes that I've implemented for my own projects. Hopefully at least some of these are useful and, if not, can be adapted to be merged into the main repository :)

1. Undefine linux if it is defined. This fixes bugs in WSL environments
2. Update ImGUI to the Docking latest version of the Docking branch
3. Remove the forcing of /MT under MSVC compilation -- causes problems with other libraries and linking
4. Set /SUMBYSTEM:WINDOWS under MSVC compilation -- Otherwise the main cinder app definition won't work as it implements WinMain (IIRC), not the default main function, causing compiler errors in MSVC

Hopefully at least some of the changes in this are useful!