There's no such thing as "optimized for Fluxbox." And the kinds of app that use GUI toolkits spend 99.99999999% of their time waiting for user input; the choice of toolkit makes no speed difference at all.
That said, I like Qt best. It's object-oriented, has
extremely good documentation, and is as fast as anything as long as you're writing a Qt app and not a KDE app. Here's an article on how much though was put into its API:
http://doc.trolltech.com/qq/qq13-apis.html
And here's a KDE program I wrote. Using Qt didn't slow it down. Writing it
in Python didn't slow it down. Making it a KDE program and not just a Qt program did make it take longer to load, but it ran at the same speed.
http://www.vcn.bc.ca/~dugan/quetzalcoatl.html
If your target platform is a Pentium 1 (speed differences between the toolkits aren't detectable otherwise), then use GTK1. It's faster
specifically because it doesn't go through Fontconfig or Cairo do to font rendering. Which means you lose support for Unicode characters and decent font rendering, you only have access to font directories specified in xorg.conf (which is no longer present on many installations), and you need to specify fonts via variations of the mile-long names you see in xfontsel and xslfonts.
Quote:
Originally Posted by slackmate
I just learnt which apps on my Fluxbox are GTK and which are Qt and I find that Qt apps look more modern with nicer icons, roundness and general eye candy. However the fonts are a bit washed out like there is too much antialiasing of as if the monitor resolition is not optimal. I compared all this with the GTK apps. The GTK applications look a bit arcane as in pre-Win XP look. However the fonts in the GTK and the general look and feel very crisp and they seem to load and respond a little bit faster.
|
These are user-adjustable settings. The defaults are
certainly not a good criteria for choosing.
See:
Quote:
I can feel a little bit of a bloat in Qt: I can see how the wingets are being built and then populated with values a split second later.
|
Have you examined that program and then ran a profiler on it? No. Then how do you know that the bottleneck is the GUI and not some other part of the program?
It is
very unlikely that you'll find even a technical speed difference between the current versions of GTK and Qt, because they both call the same sets of libraries (e.g. Fontconfig and Pango for font rendering, X to draw widgets) underneath. And no, launching a completed GTK app and then a completed Qt app doesn't tell you anything. Instead, write a Hello World program in GTK2, Qt4 and FLTK and see if you notice a speed difference.
Quote:
Originally Posted by slackmate
as well as some flickering when opening dialogs from menus.
|
Actually, Qt is specifically design to prevent this:
http://qt.nokia.com/doc/4.0/qt4-arth...uble-buffering
Quote:
The biggest problem I got with GTK is that its natively in C and its c++ wrapper might rob it off its performance advancage.
|
It will make absolutely no difference at all. Seriously, how many extra CPU cycles do you think the cost is?