Figures About SVN-Maintained Packages

For fun and profit I and gares have just harvested a bit alioth to discover how many packages are currently maintained using subversion.

The overall result is larger than I expected 3492 source packages (to be compared with 11786 total packages in unstable) are maintained using Subversion.

The numbers might be a bit approximated, I report what we did to extract these number for review.

The number of subversion-ed source packages have been computed running on alioth the script harvest source (which in turn uses svnlook its stdout can be redirected to a .txt file. Once sorted and removed of duplicates you will get something like source pkgs on svn.txt; its length approximates the number of source packages maintained via svn:

$ wc -l source_pkgs_on_svn.txt

The approximation is due to the fact that debian/control are looked for everywhere in a repository, hence it can find old packages no longer in the archive or never uploaded, and some fake package (like "MODULE NAME", but it seems to be the only one). The amount of packages which are in the above list but not in the archive is about 900 packages. Still I don't think it would be fair to exclude all of them since I guess a non trivial part of that are packages that haven't yet uploaded and hence are arguably part of the active development of Debian. Excluding them, the percentage of unstable maintained using SVN on alioth is about 22%. To that you should add the packages maintained using other version control systems.

The total number of source packages in unstable have been computed on my laptop as follows:

$ cd /var/lib/apt/lists/
$ ls *Sources
$ cat *Sources | grep ^Package: | sort -u | wc -l

The broader goal was to discover how many packages are maintained using any versioning system on alioth, but we were fluent only in svn internals. Dear lazyweb, feel free to harvest and forward to me your data for the other versioning systems available on alioth, I'll keep this post updated with news.