And yet all the time we hear how great SWT is. I've even tried writing little programs with SWT, and found it nothing more than a world of pain. I put that down to me being a Mac user, and maybe things were okay if you were on Linux, but the article suggests not.
At the same time, I and some friends have been working on a more-or-less source compatible replacement for
JTextAreamore suited for the kind of editing programmers do, and I've been really impressed with the performance you can get out of AWT/Swing.
(Especially on MS Windows! My dual G5 with a 128 MiB ATI Radeon 9600 XT is significantly slower than a random Dell box with motherboard graphics at work; a machine that feels like a dog. Where "significantly" means about 6 ms to redraw a window full of text, versus 0 ms. I should switch to
currentTimeNanos[oh, sorry, they called it
nanoTime, the damned buffoons!] to see how long the PC is actually taking. Graphics performance was something Apple regressed on in 1.4, and hopefully it will be better in 1.5 when we get it. I wonder how much of MS Windows' perceived bad performance in general is actually just because they're terrible at giving good feedback. The whole experience is so jittery that you don't notice that lots of things are actually quite fast.)
Anyway, I'll shed no tear if SWT dies before I have to worry my pretty little head about it. What I saw of the API didn't look very pleasant (unless you miss the home computers of the 1980s), the implementation sucked on Mac OS, and who wanted to install more crap to run a Java application?
Good riddance to SWT; IBM's programmers did enough damage giving Java some of its cruftiest classes; horrors like
java.text.MessageFormat. They have their work cut out to regain my trust in them, and SWT wasn't a good way to try.