Tux em Recife
Friday, October 26, 2007
Saturday, October 20, 2007
Abiword ready to maemo 4.x
Today I've finished the port of abiword[1] to new maemo SDK. In this screenshot you can see it running on scratchbox. With this new SDK is much more simple to port the currents GTK applications as the changes in maemo_app_view make this more similar and you do not need create "workarounds" :).
If you want port your application, in maemo.org[2] you can find a lot of documents to help in this task, this doc[3] is very helpful.
[1]http://www.abisource.com/
[2]http://maemo.org/development/documentation
[3]https://stage.maemo.org/svn/maemo/projects/haf/doc/hildon-window-migration-guide/appview2window.ps
Monday, October 15, 2007
Xomap 3x more fast
During these days I found this patch[1] from Dan Amelang about a Xorg optimization to work with pixbuf composite (888 over 565). This patch makes Xorg about 3x faster. I made a little change in the patch for optimization of another function too.
I've exchanged some e-mails with Dan, and apparently this patch has a problem with small images but I didn't find this problem yet. If someone would like to try this patch you can install the package[2] above and if you get the bug, please report to me. But be careful because this patch is an experimental code and can freeze your X server.
You can verify the performance with this example code[3].
[1]http://lists.freedesktop.org
[2]http://www.renatofilho.com/files/xserver-xomap_1.1.99.3-0osso37_armel.deb
[3]http://amelang.net/composite_pixbuf.c, http://www.renatofilho.com/files/composite_pixbuf
Friday, October 05, 2007
Cairo, Canvas and EXA
The last days I have used my free time to read about cairo and X acceleration. I created some tests to verify cairo performance on N800, because this is the main component for a GtkCanvas, this need be fast, cworth[1] and another cairo guys have helped me a lot in irc channel. I am still working on some tests and making some optimizations to get the best results and very soon I will publish the results.
The problem is that in N800 device the current version of cairo does not have full support for RGB 565 surfaces, and these days I read in the blog[2] from openmoko[3] guys that they have the same problems.
Another great thing is the current performance of EXA acceleration, you can see this on this benchmark[4] posted by cworth[1].
[1]http://cworth.org/
[2]http://blogs.gnome.org/thos/2007/09/26/partial-tv-fame/
[3]http://www.openmoko.org/
[4]http://wiki.x.org/wiki/ExaStatus
[5]http://cworth.org/exa/i965/render_bench/
Labels: n800 cairo Xorg EXA