On Thu, Feb 03, 2005 at 04:14:19AM -0400, bulia byak wrote:
Wow, Kees, you're a great bug fixer :)
Thanks! I'm impatient and I had today off work. Good combo for bug fixing. :)
I also spent some time cleaning up memory leaks and unassigned value usage. (Most of those cleanups were in libnr and libnrtype. Eek.) I was playing with valgrind, and I tried as many of the code paths as I could think of. It doesn't play nice with libgc, so as long as that's _INKSCAPE_GC=disable'd and I ignore those malloc's, it was a pretty interesting bit of output. I want to go run valgrind on other stuff now. :)
As for the assert-on-load, Mental is looking into it, judging by the comment in the bug. So, let's wait a bit, maybe he'll be able to fix it properly.
Yup. It could very easily be worse than I understand, so I'm happy to wait for that one to close too. I just wanted to get it back to loading enough that I could test the input extensions again.
x86_64/PPC crash bug: 1107922
sk1p verified that his bugs were gone, but we have a few other x86_64 bugs in the queue. I'd love to see if they've gone away too.
http://outflux.net/inkscape/ for anyone else doing win32 builds. There was also a problem under Linux if you had the GDK svgz loader available, it would get in the way and fail. This was fixed too.
Ted may want to look over my changes; another one was to not list file types in the load/save dialogs for a deactivated extension (one that didn't load due to missing requirements). I'd like to have them grey'd out instead (like Gimp seems to do), but there wasn't an obvious hook to grey out a ComboBoxText item.