[webkit-dev] Compile error in WebGL2RenderingContext.cpp

Salisbury, Mark mark.salisbury at hp.com
Thu Jan 17 00:27:20 PST 2019


Thanks Tim!

I didn’t know about C++ variable length arrays.  That makes perfect sense now.

Looks like people have been requesting Visual Studio add support for VLAs; there are no indications they will be supported.

From: thorton at apple.com <thorton at apple.com>
Sent: Thursday, January 17, 2019 3:50 PM
To: Salisbury, Mark <mark.salisbury at hp.com>
Cc: webkit-dev at lists.webkit.org
Subject: Re: [webkit-dev] Compile error in WebGL2RenderingContext.cpp




On Jan 16, 2019, at 20:56, Salisbury, Mark <mark.salisbury at hp.com<mailto:mark.salisbury at hp.com>> wrote:

Hello,

Hello!


I’m working on rebasing our downstream webkit port; I just pulled down recent webkit changes (since July ’18) and I hit a piece of code that’s failing to compile in Visual Studio 2017.  When I look at the code however, I’m not sure how it compiles on any compiler, or if it compiles, how it’s safe.

This is from Source/WebCore/html/canvas/WebGL2RenderingContext.cpp -

…
    int numValues = 0;
#if USE(OPENGL_ES)
    m_context->getInternalformativ(target, internalformat, GraphicsContext3D::NUM_SAMPLE_COUNTS, 1, &numValues);

    GC3Dint params[numValues];
    m_context->getInternalformativ(target, internalformat, pname, numValues, params);
#else
    // On desktop OpenGL 4.1 or below we must emulate glGetInternalformativ.

Visual Studio doesn’t like it because it won’t instantiate an array without a const size.  On the other hand, even if it compiled, passing an array with 0 size doesn’t make a lot of sense, to receive parameters back, which appears to be the intent of the code.

Why do you think it would be 0?

numValues is passed as an out arg in the first getInternalformativ call, and is presumably filled with the value for the key NUM_SAMPLE_COUNTS.

That said, this depends on a C99 feature that is maybe only available as a compiler extension in C++ and it sounds like isn't supported by the VS compiler ("variable length arrays"). So I'm guessing nobody has built it with VS yet.



The file is associated with WebGL2, but it’s added to the build in the ENABLE_WEBGL block in WebCore/CMakeLists.txt.
WEBGL2 is off by default (per WebKitFeatures.cmake) so I assume the code could not be invoked without enabling WEBGL 2.

The contents of this file are guarded with #if ENABLE(WEBGL2); you should probably figure out why that's turned on if you don't expect it.


Am I looking at unfinished code that magically compiles with gcc/llvm/etc. (non Visual Studio)?

Thanks,
Mark




_______________________________________________
webkit-dev mailing list
webkit-dev at lists.webkit.org<mailto:webkit-dev at lists.webkit.org>
https://lists.webkit.org/mailman/listinfo/webkit-dev<https://lists.webkit.org/mailman/listinfo/webkit-dev>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20190117/c677f3bd/attachment.html>


More information about the webkit-dev mailing list