On 26 May 2024 13:08 +0200, from lu...@sulweb.org (Lucio Crusca):
> Sometimes it happens some packages get automatically removed from Debian
> testing (for example because of rc-bugs), so if you ever need one of those
> packages that have been removed, you can only wait for it to be added back,
> or mix stable/unstable (or install from upstream, but that's even worse).

_Or_ you can see that an apt-get dist-upgrade / apt full-upgrade wants
to remove a package that you need, and pin the package you need
_before_ proceeding with the upgrade so as to prevent the removal of
that package, and thereby also likely causing some other packages to
be held back; and then somehow ensure that you will revisit the pin
later.

Quite a few people run Debian testing. Some even as a daily driver.
That's fine. But if you do, you need to keep in mind that it _is_
"testing". You can't expect the same level of stability as with the
stable distribution. And you definitely need to pay close attention to
what every upgrade wants to do to your system _especially_ in terms of
removals.

Mixing packages from testing and stable is no more guaranteed to work
than mixing packages from, say, bullseye and bookworm; in that it
might work, or it might not, and that the outcome will very much
depend on the specifics of the packages involved and what they use
each other for. And when it doesn't work, the failure modes aren't
necessarily going to be at all obvious.

Just because apt pinning can be wrangled into creating a system with a
mix of bookworm, trixie and sid package versions doesn't mean that
using it to do that is a good idea, or that the resulting system will
perform as expected. Apt pinning is a power tool that requires
understanding of the consequences of the results.

-- 
Michael Kjörling                     🔗 https://michael.kjorling.se
“Remember when, on the Internet, nobody cared that you were a dog?”

Reply via email to