Present Perfect

GStreamer

We all managed to squash most of the bugs we still had outstanding. I branched for release and started testing, everything works fine so far. I only had to fix a few minor issues, and I worked some more on packaging some dependencies properly.

I also finally managed to package a proper vorbis compatibility rpm.

People who want to give us a hand by testing the prerelease are welcome to do so; get either prerelease tarballs or, if you're on Red Hat 7.2 or 7.3, check out our apt repository which now also contains a test repository, which will probably contain in the future newly packaged dependencies, prereleases, and cvs snapshots.

If all goes well we can release tomorrow. After that, on to package gst-editor and gst-player properly. These days thinking up a good release name is the hardest thing of the release process, which is good.

Sad to learn that the rhythmbox guys have decided to not release in the near future and get things totally stable first. I can understand their reasoning though. It's just too bad because I keep getting people who ask for one, and even more so I've used rhythmbox from CVS for four days straight and it hasn't given one error.

I just hope the problems they seem to have internally in the code get fixed as soon as possible so people can enjoy the wonder that is rhythmbox.

gcc

We've been having issues with gcc 3.x and GStreamer, because we compile with -Werror by default. gcc 3.x has this new behaviour where it gives a warning if you add an include dir with -I that's already in the system include dir. This happens for a lot of dependencies through their pkg-config and .m4 files. I have some thoughts and questions on that and I will post them tomorrow, because I'm interested on what others say about it.

Got into an argument with hadess over the use of -Werror, he thinks -Werror is bogus. My take on it is that -Werror points out bits of code that are dodgy and ought to be properly inspected and fixed to not give a warning anymore. He reckons the warning that GStreamer had is totally bogus, and thus we should disable -Werror by default.

I don't know how it is by other people, but I found disabling -Werror just caused people to ignore warnings completely, and the times I disabled it on projects of my own led me to not find the actual error when I had one. So I'm glad -Werror is there and that we enforce it. Some of the ppc GStreamer developers seemed to just always disable -Werror and thus the warnings never got fixed.

I might be wrong about -Werror, but I think it's a valid tool, and disabling it to avoid one type of warning would cause us to miss all other (and possibly valid) warnings as well. In short : if there's a warning, either the code should be fixed or the tool that generates the warning should be fixed.

On the plus side, David Schleef just commited a fix for the cause of the warning, and gave a good explanation of why it generated the warning and what could happen because of it. I feel strangely vindicated ;)

GStreamer We all managed to squash most of the bugs we still had outstanding. I branched for release and started testing, everything works fine so far. I only had to...