Hello,

I'd like to raise another issue I've met again recently. Shortly put,
some of our projects are relying too much on their overlays. The net
result is that some of their packages in the tree are not well-tested,
semi-broken and users end up being hurt by that.

The major project where this can be seen is science. With no offense
intended, but I'm afraid that sometimes the team itself is losing track
of what has been committed to the tree and what is in the overlay,
and especially which versions are compatible.

Another similar project having this problem seems to be lisp. From bug
#465864 (which points to many other bugs not fixed in gx86), you can
gather:

  "Anybody who intends to use something lisp-related (like maxima)
  in Gentoo seriously always uses this overlay. There are too few
  developers in the common-lisp herd, and the main tree remains
  neglected for years." (by Andrey Grozin)

which shortly shows that in some areas the issues are really serious.

Teams, what are the main reasons for keeping that much stuff
in overlays? What can be done to avoid it?

While I can see the benefits of, say, testing extraordinarily
experimental stuff in overlays or keeping there stuff that is not
intended to land in gx86 at all (like some custom hacks), I feel like
just keeping the newer versions of some packages is more of issue
breeder to us.

Please remember that most of our users doesn't know those rules.
If I am looking for a good mathematics package, I take maxima, though
I have almost no idea of lisp except for parentheses. The lisp-related
flags are confusing to me and ever worse is the fact that the default
choice simply doesn't build. Then I try alternate implementations.

Expecting users to grep bugzie or some other kind of pages to find that
they are supported to install an overlay to properly use package that
is in gx86 is not good. The sole existence and use of overlay is
causing the gx86 package and/or its deps to be in increasingly worse
shape.

If the problem is really manpower, I think you should try to work with
proxy-maint. If that's not enough, then we need to find a better
solution.

In the worst case, we may prefer to move some of the packages out of
gx86 and specifically expect all users to use an overlay, consistently.
But in this case, we should probably consider redesigning Gentoo to be
based more on official or semi-official repositories like Exherbo
so that all users would have equal rights.

As a last note, I'd like to note that I'm talking about lisp that much
because maxima is a recent case where I've seen this. But there were
even worse things with science overlay, lapack and blas -- including
getting the system into a state where neither gx86, nor science overlay
packages work.

-- 
Best regards,
Michał Górny