Although a bit of old news now but for those who haven't heard, GUPnP's first feature complete release is out. Since it now provides the same level of API that intel's libupnp provides, all the projects out there that were using libupnp until now can (and must) start using gupnp now if they want to be gmainloop-friendly and don't want any threads to be launched behind their backs.
To be (a git) or not to be
I tried git for the first time some two years ago when all i wanted to do was to grab a branch of linux source tree. That experience was enough for me to be really scared of using git ever again.
A month back, I ran into a typical problem in centralised SCMs when a contributer doesn't have write access to the repo: I sent a patch of my changes to Jorn and he committed the patch with some changes but before he actually committed it, I had made further changes on the bases of my previous changes so when i updated my working dir, SVN was able to detect conflicts but conflict report wasn't a very correct one and quite confusing to say the least. So despite my efforts, I ended-up overwriting Jorn's changes.
Matthew Allum advised me to use git-svn to avoid this problem in future but I just didn't want to use git again until i had a beer with Timo Savola a week back. He told me that the user-interface of git has improved quite a lot over the last few years and he sees no reason anyone should use any other SCM anymore. So on his advice I played with git and git-svn for the whole day on Friday. It was quite a frustrating experience at first but in the end when I got a hold of the basics, I realized that I was thinking in terms of other SCMs. So now I am also a proud supporter of git and what convinced me were strong arguments from Timo and this presentation by Linus himself (which was also pointed out by Timo).