[aur-general] Is it okay to mark broken packages out-of-date on AUR?

[Peter Simons]

Hi guys,

the AUR user palmfron has recently flagged the package "haskell-haskcore"
out-of-date, because the PKGBUILD is broken. It cannot be compiled,
because it depends on other packages that no longer exist:

  http://aur.archlinux.org/packages.php?ID=20383

Now, there is disagreement among the members of the ArchHaskell team about
whether it's okay to flag that package out-of-date. Some argue that the
package is *not* out-of-date, because the published version 0.1.0.4 is the
latest one available, so the package cannot be updated to a newer version.
These people argue that flagging a package out-of-date just because it's
broken is not alright.

Others say that it's perfectly alright to flag that package out-of-date,
because it's *broken*, so clearly the PKGBUILD does need updating to be
useful.

Is there some sort of consensus among AUR maintainers how to deal with
that kind of situation? If an AUR package is current, so to speak, but it
doesn't compile, then what should be done with it?

This issue is of some importance for us, because the 'arch-haskell' user
has published an approximated 500 packages on AUR that are broken, i.e.
these packages cannot be built because of unsatisfiable dependencies:

  https://github.com/archhaskell/habs/issues#issue/4

I'd appreciate any advice that you could offer.

Take care,
Peter