Hello Sven, On Thu 09 Sep 2021 at 01:08PM +02, Sven Bartscher wrote: > I was just looking at the reason why hgettext is marked for removal from > testing. While doing so, I noticed that the last releases of the package > have all been bumps to the dependencies. The last upstream release is > from 2017, so apparently it is time to consider the package unmaintained > upstream. > > I think the original reason hgettext was packaged, was because I cared > about it. This has changed, I don't use the package anymore and don't > see a reason to have it in Debian. So from my perspective the package > can be removed from the archive. > > While checking the package-plan to see if hgettext can be safely > removed, I noticed that it outputs 129 packages which aren't key > packages or needed by key packages, i.e. candidates for removal from the > archive. See below for the whole list. > > Since I haven't been active in the Haskell Team for a long time, I'm not > sure what the current practice is. But when I was still around, we tried > to keep the set of packages small, by removing packages that nobody > cares about (as identified by the “key” mark in the package-plan). Thank you for looking into this. > I don't want to suddenly start removing packages from under peoples > noses. So maybe someone can clarify: Do we, as a team, still try to > remove non-key packages or has the consensus shifted to just maintaining > everything forever? If the former, maybe people should take a look at > the packages in question, decide if they still want them in Debian and > mark them as key packages (preferably with a comment explaining who to > poke when the package breaks). Then we could set a deadline at which the > remaining packages can be removed. Ilias has taken the tack of filing a bug against the package indicating intent to remove, and then after a certain number of days converting that bug into an RM bug. In most cases no-one said they wanted to keep the package. I think this is reasonable given how each package can impact the workload of anyone trying to update anything Haskell at all. So, no change in team consensus. > There are some packages in the list, which should probably not be > removed. For example I think cabal-install should stay in Debian, even > though I don't personally care about it. Some other packages look like > there might be actual users who care about them (such as propellor, > xmobar, xmonad-wallpaper). So maybe they should stay too? We could set aside packages building binaries for now, as they are far more likely to be in use by end users, and there are far fewer of them? A few other examples from the list are pandoc-citeproc-preamble (which I know people are using to build documents, including me) and debug-me. -- Sean Whitton
Attachment:
signature.asc
Description: PGP signature