Re: Colour Corruption on remote X11 with different endians
Hi David, could you please add this to our bug tracker with as much details (eg versions of libraries) as possible. Perhaps even screen shots of different standard windows with the different backends. I don't have a ppc machine available, so I am not able to test any of this myself. Still I would like to help resolve the issues. When you write that with cairo everything has its colour inverted, does this include lines, fills and text? I am sure that bitmaps get treated wrongly by our current code, but the other stuff surprises me. As for the current art code, I knew that depths of 16 were handled incorrectly, but this may be the first report that 32 bit get handled wrongly as well. Fred David Chisnall wrote: I've done a bit more testing with GNUstep and remote X11 with the X server running on PowerPC (little endian) and the apps running on x86 (big endian): With the xlib back end, everything is fine. With the art back end, almost everything is fine, but pixmaps have their colours inverted. With the cairo back end, everything has its colour inverted. Since the xlib back end lacks a number of features, this isn't ideal. It is also, however, a lot more responsive than the other two (with art being slower than cairo). ___ Gnustep-dev mailing list Gnustep-dev@gnu.org http://lists.gnu.org/mailman/listinfo/gnustep-dev
Re: Colour Corruption on remote X11 with different endians
I've wondered for a long time why I always get red text as blue and sometimes also black text as blue. I never realized it could be due to endianness differences. I use the same combination as you, GNUstep on x64 (-64), X server on PPC, art backend. Tim On 16 Aug 2008, at 22:39, David Chisnall wrote: Hi Everyone, I've done a bit more testing with GNUstep and remote X11 with the X server running on PowerPC (little endian) and the apps running on x86 (big endian): With the xlib back end, everything is fine. With the art back end, almost everything is fine, but pixmaps have their colours inverted. With the cairo back end, everything has its colour inverted. Since the xlib back end lacks a number of features, this isn't ideal. It is also, however, a lot more responsive than the other two (with art being slower than cairo). David ___ Gnustep-dev mailing list Gnustep-dev@gnu.org http://lists.gnu.org/mailman/listinfo/gnustep-dev
Colour Corruption on remote X11 with different endians
Hi Everyone, I've done a bit more testing with GNUstep and remote X11 with the X server running on PowerPC (little endian) and the apps running on x86 (big endian): With the xlib back end, everything is fine. With the art back end, almost everything is fine, but pixmaps have their colours inverted. With the cairo back end, everything has its colour inverted. Since the xlib back end lacks a number of features, this isn't ideal. It is also, however, a lot more responsive than the other two (with art being slower than cairo). David ___ Gnustep-dev mailing list Gnustep-dev@gnu.org http://lists.gnu.org/mailman/listinfo/gnustep-dev
Re: Colour Corruption on remote X11 with different endians
Hi, welcome to the club. On 2008-08-16 23:39:44 +0200 David Chisnall [EMAIL PROTECTED] wrote: Hi Everyone, I've done a bit more testing with GNUstep and remote X11 with the X server running on PowerPC (little endian) and the apps running on x86 (big endian): I did it the other way. With the xlib back end, everything is fine. With the art back end, almost everything is fine, but pixmaps have their colours inverted. With the cairo back end, everything has its colour inverted. xlib works fine for me too, art used to work, cairo is broken, but it is not inverted it has a strange color permutation. (everything is bluish) Since the xlib back end lacks a number of features, this isn't ideal. It is also, however, a lot more responsive than the other two (with art being slower than cairo). Ideally, they should all work. Xlib is by way faster (if you think it is smarter, it doesn';t just shovel bitmaps over the network but, for example, uses the server to display the fonts) especially if you use fonts with bitmaps and not AA. Then Menus look like local. Scrolling is slow though. just as a test I did here x86-ppc with art and the corruption is strange: some elements adre correct in color, some nots (blue cast). But for example clicking on the scrollers makes them alternatively display in the correct or in the wrong color, while test in the menus displays black and gets blue when clicking on it... I'm almost sure it used to work years ago. Riccardo ___ Gnustep-dev mailing list Gnustep-dev@gnu.org http://lists.gnu.org/mailman/listinfo/gnustep-dev