Yesterday i just installed the new Eclipse version at work and suddenly i realized i could not use up/down arrows to use lineedit's history. I filed a bug against their bugzilla that got closed as duplicate of a WONTFIX/NOTOURBUG bug :-(
Actually it seems the bug is on gtk but it quite astonishes me the fact that i'm using the same gtk i was and it worked on previous version of Eclipse.
So that forces us to ask the question: As a software developer, it's better to release a product that does not work and wait for the library we use to fix itself or should we workaround the bug as much as we can?
I'm not sure which one i'd pick with my developer hat, but as a user i tell you i want things to work :D
the swt download contains the compiled libs for the target architecture. i'm guessing that the eclipse linux download also bundles it's own compiled gtk libs so that's probably the problem (you're actually _not_ using the same libs).
ReplyDeleteSure, users want things to work. But this is about longterm-vs-shortterm. On the long run, improving the lib would be best. If you can't, waiting would still provide more incentive for the lib devs to fix it. Last, you'll be doing work which is useless on the long run.
ReplyDeleteBut yes, in short-term, you improve the app. Until the bug gets fixed, of course.
@anonymous: i did an ldd of the swt libs and they seem to be using my local gtk libs
ReplyDeleteIt's open source. Just go fix the lib.
ReplyDelete@anonymous: Good try, i'll put it on my long todo list, maybe will be done by 2014 or so.
ReplyDeleteThis remembers me...
ReplyDeletehttp://bugs.kde.org/show_bug.cgi?id=125969
:-)
@Pedro: This is a radically different issue, subclassing a widget and implementing the functionality should be around 10 lines of code, fixing a bug in freetype from inside kpdf is impossible.
ReplyDelete