Debian Haskell packaging team getting underway
It is reasonably well known amongst Haskell enthusiasts who use Debian/Ubuntu that the experience of using and developing with Haskell on these distributions is, to put it diplomatically, not as good as some others are able to provide (Arch, I'm looking at you, you sexy heroes). A couple of recent mails to debian-haskell have made this abundantly clear to us; we have a bad situation and blocking on maintainers (especially since some of them appear to be inactive) is just making things worse.
Enter the Debian Haskell Packaging Team! This had been proposed previously, but never carried forward for various reasons – including, importantly, nobody just rolling up their sleeves and getting it done. Recently John Goerzen wrote a mail which shone a bright light on several areas of our Haskell infrastructure that need work. This seems to have spurred several people into action, and John took the necessary administrative steps to get the pkg-haskell team back up on its feet (it did indeed exist previously but seems to have fallen into disrepair with only two members). As I write, there are now several packages under team maintainership:
laney-guest@alioth:/git/pkg-haskell$ ls
agda.git haskell-convertible.git haskell-ghc-paths.git hdbc.git
gtk2hs.git haskell-ghc-mtl.git haskell-x11.git
laney-guest@alioth:/darcs/pkg-haskell$ ls
_darcs haskell-ghc-mtl haskell-haskeline haskell-ifelse haskell-terminfo
gtk2hs haskell-ghc-paths haskell-hint haskell-monadcatchio-mtl haskell-zlib
We truly live in exciting times (if you care about this sort of thing). Now, as soon as the debates about group policy are over, and we've decided what VCS to use (urgh…), we should be able to get down to some actual work.
But why will this help? The main areas we have problems with are:
- Core infrastructure packages (outside of the GHC complier) being out of date as they are maintained by inactive developers.
- Uninstallability caused by new upstream releases of GHC. GHC releases make no ABI compatibility guarantees so with each new upstream version we need to rebuild the entire dependency chain. This takes time and can not always be done through binNMUs. Inactive maintainers can again block here.
Being able to grind through all packages and fix/upload on new GHC releases will provide a massive improvement to the churn that is currently inevitable after an upstream release. And being able to update out of date packages is an obvious win for team maintainership.
Once we have these issues cleaned up, we can move on to the real goal of implementing the Haskell Platform in Debian. We should now be better prepared to cope with the increase in package load that this will necessitate. Contributors, too, in seeing more cooperation should feel more inclined to give their time.
If you're reading this and would like to get involved with the team, helping us to maintain existing packages and to package up new libraries then join us at #debian-haskell on OFTC. We'd love to see you along.
Here's to a happy future!