Re: pkgsrc-2024Q3 net/tigervnc is broken on sparc64

[Sad Clouds <cryintothebluesky%gmail.com@localhost>]

On Sun, 27 Oct 2024 12:42:57 -0400
Greg Troxel <gdt%lexort.com@localhost> wrote:

> Have you filed a bug report with upstream?  You haven't included the
> URL.
> 
> If no one wants to spend time fixing the code, it isn't going to get
> fixed.  That's where we are.

Not yet, because from personal experience bug reports can linger for
years without anyone paying much attention to them. What tends to
happen, sometimes - if it works on x86 Linux, then it is good enough
for us and who cares about vintage architectures like SPARC.

So at the very least, I would need to build the package with debug
symbols, get backtraces from core dumps, see if I can fix the issue,
then rinse and repeat, as there may be multiple issues. This will take
some time. And in the meantime, it is far simpler to roll back to the
previous working version.

You are correct it is not a pkgsrc bug, but pkgsrc is meant to be used
as a package management system, rather than some glorified wrapper
around third party software. So pkgsrc has facilities to include its own
patches, for example, if upstream are not fixing the issue. It should
also include facilities to deal with packages that develop security
vulnerabilities or serious regressions. I'm not trying to dictate a
policy here, but one way to deal with such issues is to revert to a
previous working version, backport important fixes from head to a
stable branch, etc.

At the moment, the situation seems to be, if a new stable pkgsrc
release includes a broken package, then you should:

a) Wait for the next several months or years, until someone has the
inclination to fix it.

or

b) Start digging through previous revisions and manually try to revert
to an older version. This is not ideal for anyone installing prebuilt
binary packages.

Perhaps there needs to be a fallback process, where a stable pkgsrc
branch can either downgrade a package, or include a second copy of it
under a different version, until all regressions are fixed. It's just
an idea, I'm not really forcing anyone to follow it.