Kein-Hong Man wrote:
Robert Roessler wrote:
Neil Hodgson wrote:
Robert Roessler:

2) is the Pango/GTK on Windows combo doomed?

   Do you read Federico Mena Quintero?
http://primates.ximian.com/~federico/news-2005-10.html
   Don't know if any of the optimizations have made it to a GTK+
release or onto the Windows port.

A current summary of my (superficial?) performance investigations:
[snip]

I thought this was normal (that Win32 GTK+ apps have a different kind of responsiveness.) The whole GTK+ kaboodle is a pretty large library, and it is another layer sitting on top of Win32. Redraw definitely behaves differently compared to Win32 native apps. Try this on any big GTK+ app on Win32: GIMP and Inkscape are all a little less than crisply responsive, and processor usage usually spike quite high when large redraws are needed. Scrolling say, an Inkscape tutorial causes processor usage to shoot up pretty easily.

I sometimes wonder about this headlong rush to get out from under the thumb of the Evil Empire... ;) One of the costs seems like we are not getting to have as much [peppy] interface fun and GUI goodness as we should, even though processors, memory, graphics, and disks are faster than ever... I am still bummed about SeaMonkey [Windows] builds quitting using GdiPlus around 10 weeks ago - while I do not have a good sense of what their rendering architecture looks like now, I do think that screen update performance took a nose-dive at that point.

Yeah, GTK+ on Win32 probably can do with some more optimization, but I see this as primarily a GTK+ on Win32 issue, not a real problem with Scintilla/SciTE. Improving GTK+ on Win32 will benefit all apps that use it; modifying Scintilla/SciTE seems like a short-term fix.

Philosophically, your point is hard to argue against... OTOH, I have some clues how Scintilla works (no comments from Neil), so the problem is a bit more tractable. But I have pretty much exhausted approaches based on Scintilla enhancements... not that there aren't any to be made, but I think any inefficiencies in Scintilla text rendering are dwarfed by the GTK+ on Win32 implementation.

But then that is how the performance-oriented optimization game works - you take care of the worst offender(s), and then next tackle the former #2 or #3 performance-sapping areas of code (which is exactly what the account of the Pango speed enhancements talks about). :)

Robert Roessler
[EMAIL PROTECTED]
http://www.rftp.com
_______________________________________________
Scintilla-interest mailing list
[email protected]
http://mailman.lyra.org/mailman/listinfo/scintilla-interest

Reply via email to