Hi Andrew, This sounds like a solid plan. For 5.6, we probably don’t have any other choice than staying with what we have now.
For 5.7, the desktop this sounds ok IMO. I actually don’t think we need to (or should) support VS2012 there at all. I’d only like to get a confirmation from the WEC maintainers, that they are ok with this plan as well. Cheers, Lars On 21/09/15 11:50, "Andrew Knight" <[email protected]> wrote: >Hello, > >tl;dr: I propose that the copy of ANGLE currently in the 5.6 branch of >qtbase be maintained without being upgraded for the next minor release >of Qt (5.6). For Qt 5.7 and beyond, we should upgrade ANGLE from >upstream while dropping support for VS2010 and VS2012. > >Around each minor Qt release someone (usually me) updates the >third-party ANGLE sources in qtbase. The last time this happened was for >Qt 5.5, imported from Chromium branch 2356, dated 01 April 2015. > >Since then (and including a good chunk of code up to that commit), the >Chromium project has adopted C++11 and C99 usage in its codebase (due to >switching to VS2013 as the minimum Microsoft compiler). That means a >great deal of code no longer builds with older VS versions, and we >already patched ANGLE for VS2010/VS2012 in Qt 5.5 to deal with some of >these issues. Due to the invasiveness and backward thinking of such >patches, they will not be accepted upstream. > >Since then, the amount of such required changes to upstream ANGLE has >exploded; the upkeep required here is truly non-trivial. Therefore, I >now propose that we *do not* upgrade ANGLE for 5.6. Post-5.6, we will no >longer apply patches to support older compilers. The version in Qt 5.6 >will be maintained as any other third-party library in Qt (receiving >patches as needed), while Qt 5.7 will only support VS2013 and newer (as >well as recent versions of MinGW, but that is tangential to this >discussion). > >As Qt 5.7 will already drop support for VS2010, I believe the only >supported compiler this affects is VS2012. This compiler is supported >for Windows Embedded purposes, a platform which does not use ANGLE >anyway (see opengl.prf for details; a WinCE SDK normally provides its >own GL stack). For keeping VS2012 Desktop binaries building, we will >need to forcefully disable ANGLE for that compiler. I think this is a >fair trade-off, and one which has already has a precedent set by Qt >WebEngine. > >Now is the time to speak up (I'm looking at you, Windows and WinCE >maintainers) if there is something wrong with the above proposal. Of >course, inaction will probably lead to the same result: ANGLE will not >be upgraded for 5.6, and we can revisit the issue nearer to the 5.7 >release when it's time to upgrade again. > >That said, any ANGLE-specific bugs can be assigned to me and I will do >my best to fix them or back-port the relevant changes from upstream as >needed. > >Cheers, >Andrew >_______________________________________________ >Development mailing list >[email protected] >http://lists.qt-project.org/mailman/listinfo/development _______________________________________________ Development mailing list [email protected] http://lists.qt-project.org/mailman/listinfo/development
