Hi Emil, On 16 February 2018 at 10:40, Emil Velikov <[email protected]> wrote: > On 15 February 2018 at 23:12, Derek Foreman <[email protected]> wrote: >> Maybe it's just me, but having different names for the same variable for C >> vs C++ compilation seems really nasty and potentially surprising. >> > Indeed it is surprising. In practise nobody should be using C++. > The guards are added just for lolz - but we can flip things the other > way around. > Namely, error out when using C++ ;-)
I mean, effectively we did before the rename, by using a reserved keyword ... >> Shouldn't old versions of Mesa already have internal versions of this header >> anyway? I didn't think it was publicly installed until now... >> > That's what Dan't patch did. Sadly as-is (before this series) every > Mesa release under the sun - released or git will fail to build. > I don't think we can retroactively go and fix the all :-( Every one? As far as I'm aware, it's only for 17.3.x. If we apply my patch to explicitly pick the path, then all 18.x will be fine (since they use an explicitly internal path), 17.2.x and below are all fine (since the file was still called wayland-egl-priv.h), and presumably 17.3.x will be seeing another stable release which could trivially fix that as well. OTOH, if we keep releasing Mesa with ambiguous includes of wayland-egl-backend.h, then we can't change API details inside that struct, for as long as we want to support these releases. Hence why I'd go with the patch to disambiguate the includes, or maybe renaming wayland-egl-backend.h in Mesa to wayland-egl-mesa-backend.h or something. Cheers, Daniel _______________________________________________ wayland-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/wayland-devel
