Ian Ward wrote: > Nuutti Kotivuori wrote: >> Thanks! I tried the patch and it alleviates the situation a >> lot. However, performance is still not good enough. Here is a sample >> program to check the performance (not minimal, one of my test files): >> > ... >> When run in ISO-8859-1 mode, performance is great - when run in UTF-8 >> mode, even with the patch, I can easily max out the CPU and have the >> display updating really slow by rapidly tapping left and right. > > I've updated the patch. Give this version a try and let me know.
This updated version works like a charm! The CPU usage when switching columns is mostly identical with ISO-8859-1 and UTF-8. I tried a couple other test programs as well and the speed is great there as well. The difference is also noted when running the graph example from urwid distribution - it works nice and smooth with a large terminal now. There is only one case where I could get a noticeable slowdown - and that was an edit field where I wrote around 20 lines of the japanese character 'a'. After that, every keypress took like a second to be processed. But the only thing alleviating that would be to actually make the calc_width calculation fast somehow. Pyrex or calling the system wcwidth or something. Anyway, thanks a lot for the fixes - I will be mailing more stuff here as it comes up. -- Naked _______________________________________________ Urwid mailing list [email protected] http://lists.excess.org/mailman/listinfo/urwid
