Hi Chris,
On Tue, Jan 14, 2025 at 07:20:33PM -0500, Chris Frey wrote:
I won't speak for Tobias, but I'm guessing he was waiting on more news
from me, and I got sick and dropped the ball.
Sorry to hear that.
What costs do broken packages incur? Is it the FTBFS bugs?
It is not as boolean as it may look. There are very many QA teams in and
around Debian that perform archive-wide work and the mere presence
incurs a small cost on each of them. We generally consider that fine. Practically speaking, the FTBFS is a pretty high cost to many of them.
It makes Lucas/Santiago (who do dedicated FTBFS tests) look into your
package. It also makes the reproducible team look into your package.
Many transitions (such as R³-default or /usr-move) use rebuilds to judge
their effects on the archive and a package that fails to build becomes
an unknown datapoint and extra effort.
If we get that fixed, will we be less of a burden?
A lot, but then the package still cannot migrate to testing. With rare exceptions, we should only keep packages in unstable that are meant to
be part of stable. Unless there is some way of eventually fixing the
missing source bug, we should get rid of it. If all else fails, moving
fava to non-free could be an option (and there it could actually
migrate).
Fava is a very useful package, and I hope to see it stay in Debian.
I hear that. Can we find a compromise? If the FTBFS gets fixed and some
kind of plan for fixing the missing source one (e.g. moving to non-free
or detailing how the source shall be included) is sent to the bug
report, I am convinced that the cost that fava incurs on Debian is
warranted. It was the combination of serious problems with an extended
period of lack of progress that made me suggest it for removal. If you demonstrate progress, that argument goes away.
If the FTBFS fix is simple (no new upstream release), I offer sponsoring
the upload.
Helmut
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)