Should packages known to not work at all be in the repo?

[alexeicolin]

What does the community prefer: should packages known to not function at all be present in the binary repositry and in the master branch of the PKGBUILD repository?

If a package is in the repo and if there are no issues open against it in the bug tracker, the user assumes that the package should be more-or-less functional. And, when major problems arise, the user suspects that it is a problem in his/her environment.

For example, the ghc package is non-functional at all at this point in time (see the closed Issue #1062). But, it is installable from the repo and present in the master branch. And, most problematic, bug reports against it are dismissed instead of used to track outstanding issues. The rationale given is "The packages only remain to potentially help building in the future."

A more appropriate place for packages that exist "to potentially help building in the future" is a development branch, not the master branch. If the package is in the master branch, then it would be much more helpful to the users if bug reports against it were kept to track outstanding issues with the packages that come from the official repository until the issue is invalidated, fixed, or determined to belong upstream. Does the leadership disagree that this strategy would be more helpful to the users?