Programming

I maintain a fair bit of Free Software and make releases on a fairly regular basis. Automating the build process is rather simple these days, with tools like make being pretty ubiquitous. I'm a Debian developer, and so all my software gets uploaded there, and that process is automated too.

But there's still a lot more to do. The process from there generally looks like this:

1. scp the files for the release to the home web (or gopher) site for the package


2. ssh to the web server and:
   
   
   
   1. Delete or move files for the old release
   
   
   2. Extract updated documentation from the new release so people can read it online
   
   


3. Post to the relevant mailing list with info about the new release (and a standard bolierplate of background info


4. Log on to Freshmeat, go the release area, fill out several forms, and announce it to freshmeat.

People that use SourceForge or another *forge have it even worse, having to log on to yet another site and fill out a bunch more forms. (In my process, I don't have to use a browser till I get to Freshmeat).

I think we need some better tools here. We need to have a server-side set of tools for managing software, while not being so massive or difficult as SourceForge. On the client side, we need scriptable tools that can do all of the above automatically, pulling out relevant version numbers and change messages from metadata in the tree (such as debian/changelog or SPEC files) or the directory name itself.

If I get some time, maybe I'll write such a thing myself. In the meantime, maybe someone else will get to it first. I hope so.