Re: 2024Q4 gif2png build fails

[Greg Troxel <gdt%lexort.com@localhost>]

Taylor R Campbell <riastradh%NetBSD.org@localhost> writes:

>> This would be an option. Alternatively, I could backport the fix.
>
> Patch looks pretty straightforward.  Would have been nice to backport
> it before the branch, but since the sysctl and /proc vectors both
> reach the same data, I think this is reasonably low-risk to backport
> even to the 2024Q4 branch.

Speaking from experience of handling freezes, there are two separate
issues.  One is posting the fix to others (in pkgsrc-current) and having
an opportunity for testing befoer a pullup request.  That is of course
normal.

The second is trying to do that before an impending branch cut, which
has elevated risk from doing work under deadline pressure, not having
enough time for other to test, bulk builds needing to redo, and just the
n^2 problem of risk from changes.  I don't really mean n^2, but when 4
things change and then you have to figure out which did what, it's
vastly harder than 4x checking on single things.

With a pullup, the builds will be ok, and then it will be clear if the
pullup is trouble.  But with doing this at normal pace, I think the risk
is tiny.

Just explaining why "get this in before the freeze is over" isn't always
a great plan, even if it's the outcome you want.