-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Am 2013-12-11 14:32, schrieb Henri Verbeet:
Since commit 6bc69c271bf440b4ecad012dc57888fdffc075d5, the GDI implementation assumes DIB sections always exist.
Actually the idea behind this patch was that all paths that need a DIB section create one if necessary. Is there a place where that's not true? If it's about x11_copy_to_screen I think the better fix is calling surface_getdc from there.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Am 2013-12-11 15:49, schrieb Stefan Dösinger:
Actually the idea behind this patch was that all paths that need a DIB section create one if necessary. Is there a place where that's not true? If it's about x11_copy_to_screen I think the better fix is calling surface_getdc from there.
Actually, that would be inefficient due to SFLAG_PIN_SYSMEM, so your patch makes perfect sense. Still I'm curious what the exact problem is.
On 11 December 2013 15:57, Stefan Dösinger stefandoesinger@gmail.com wrote:
Am 2013-12-11 15:49, schrieb Stefan Dösinger:
Actually the idea behind this patch was that all paths that need a DIB section create one if necessary. Is there a place where that's not true? If it's about x11_copy_to_screen I think the better fix is calling surface_getdc from there.
Actually, that would be inefficient due to SFLAG_PIN_SYSMEM, so your patch makes perfect sense. Still I'm curious what the exact problem is.
Yes, it's x11_copy_to_screen() in particular that triggered this, although it may not be the only place.