I’m not sure, maybe I already used this title before, as this is something I do often. I don’t like warnings and I always try to fix them when I can.
The reasons of this are that I think that every warning is just an error in disguise unless it’s properly handled, and that often when you look for a runtime error you might most likely find it looking at warnings than trying to run gdb over the whole code.
So today, as I was feeling pretty tired of doing the usual things, I decided to clean up warnings, from xine-lib this time. I did that before actually, but there was too many rejects in the CVS so I dropped the idea of fixing them one in a row. Now that I have CVS access I can fix them directly there without having to deal with rejects in the next release.
Now, I removed the most important or annoying bugs from xine-lib (format string bugs are annoying because they might mask other kind of format warnings), I also found a couple of logic errors with comparisons between signed and unsigned that could have caused some params to be ignored, or a loop to not end properly and cause an invalid access.
Next step is to make xine-ui have less warnings, too, so I’ll be working on that later. I’ll see to prepare a CVS snapshot version of it for Gentoo, but that will remain package.masked unless someone finds a reason for it not to be, as it was for xine-lib originally. I’ll be asking a new release of that soon anyway, that unmasked, as there are a few security issue that were fixed in the cod but not yet released.
Oh, thanks to leio, I now know that the R200-based ATI cards works fine with MergedFB (rather than xinerama, but that’s the same to me), so I’ll be planning on taking the 9250SE card for €55…