On 2/12/07, John Zelle <[EMAIL PROTECTED]> wrote:
Kirby (and others), I've had a number of questions lately about the slowness of my graphics library. I've noticed that some programs run well under Linux (my development platform) but run very sluggishly under Windows. Was your testing on Windows? Just curious. If you do get a chance to try your code on both Linux and Windows, I'd like to know if you notice much difference. When I get some time (won't hapen until summer) I might have to dig in and try to address these performance issues on non-Linux platforms. If anyone has any thoughts on the source of this discrepancy, I'd love to hear them. --John
Yeah, right on. I was using WinXP for the nks.py + canvas2.py + graphics.py test, and getting every Rectangle triggering a canvas update, thereby slowing the rasterization. I copied the same .py files to my Ubuntu box, and there's a delay, while the entire canvas is populated behind the scenes, and then the Mayan Pyramid pops up, already complete, just as it should. Gregor's straight Tk version in canvas3.py doesn't update per Rectangle on WinXP, when I comment out the canvas.update line per his notes. I tried for like an hour to figure out where the canvas refresh was occuring from your graphics.py and couldn't track it. Your __autoflush method looks good and I could detect no calls to Tk's root.update(). Appears to be a thorny problem. Kirby
_______________________________________________ Edu-sig mailing list [email protected] http://mail.python.org/mailman/listinfo/edu-sig
