Translators need you; Lokalize brokenjdbrx
Lokalize is the translation tool (not only) for KDE. So if you use KDE in another language than US-English, chances are good that someone used Lokalize to enable you to. But Lokalize is sick now and needs some medicine.
Actually it is not Lokalize itself that is sick but some part of Qt. But let me start at the beginning.
A few years ago we had a problem with KBabel, the former (not only) KDE translation tool. It appended some garbage to random strings when switching to the next string to edit. This led to faulty translations in all languages. But what does KBabel have to do with Lokalize? Well, history repeats itself.
A few months ago, Lokalize started to crash when backspace was pressed at the end of a string. This bacame very annoying and gave me a lazy finger for the backspace key. By now I do not use backspace at all anymore.
It turned out, Qt was the culprit. A regression in 4.7.1. And there was an upstream commit that seemed to be fixing the issue. So we went to our distro’s packagers and requested the inclusion of the fix into the 4.7.1 packages. Problem solved.
A few days later this strange appended garbage in random strings returned. First I thought, we became careless but when the incidents heaped up, the backspace crash fix came to my mind, and indeed, others experienced the same issue.
There are reports upstream. One from our ranks which is in “needs more info” state asking for a minimal example that demonstrates the issue and another one from a Psi developer who provided such an example but which is closed due to the fixed crash.
So what now? Developers to the rescue please. Help making KDE rock in a polyglot way. :)
If you know how to read Qt code, please have a look at the code and if you figure out, how to fix the issue properly (ask for testers in the bug report), please send the patch to the Qt bugtracker. We will try to get the fix into our distro’s 4.7.1 packages as well.
And for the translators: remain at Qt 4.7.0, if you can. It’s safe there. :)
we do not only need a fix, also a unit-test if possible.
I think we should just revert the performance optimization which most likely caused this regression:
http://qt.gitorious.org/qt/qt/commit/f62f6effab8d1551d8e5e5843dc478addee96de1
and the followup:
http://qt.gitorious.org/qt/qt/commit/3cf6e87b17d6ca722eea804d8e758879836f63fd
Using Qt 4.8 here from openSUSE’s repos: No crashes. Everything is smooth.
thanks for brinning this up. i left a comment in http://bugreports.qt.nokia.com/browse/QTBUG-15857 about the probable fix-for-the-fix
@Kevin: Well, I would prefer a solution that comes from upstream. Our current solution might even be half of the solution worked-out by upstream. Noone know if we missed another commit that makes the fix we found working.
@Markus: What’s Qt 4.8? :) I think it’s not decided yet, what master will become in the future.
@Nick: Thanks for looking into it. Are you able to provide a minimal Qt example as requested yesterday in the report?