[csw-pkgsubmissions] newpkgs cups, cups_devel, cupsclient, cupsd, (...)

[Philip Brown]

On 12/22/10, Maciej Blizinski <maciej at opencsw.org> wrote:
> Upate of cups packages.  Fixes bug 4634, "CSWcupsd 1.4.3,REV=2010.06.29 no
> longer starts"[1].  This update is special in that it contains unoptimized
> binaries with debugging symbols.  cupsd-1.4.5 built with optimization and
> ran
> in my test environment segfaulted when reading configuration:
>
> (dbx) run -f
> Running: cupsd -f
> (process id 11919)
> Reading libc_psr.so.1
> Reading en_US.UTF-8.so.3
> Reading methods_unicode.so.3
> t at 1 (l at 1) signal SEGV (no mapping at the fault address) in
> cupsdReadConfiguration at 0x27894
> 0x00027894: cupsdReadConfiguration+0x24fc: ld [%o5], %i4
>
> When built with debugging symbols, it would work without segfaulting.  Not
> being able to debug the issue in a reasonably short time, I'm inclined to
> release unoptimized - but working - binaries, to alleviate the current
> problem.


errr.. as a programmer, this screams to me, "STILL BROKEN YOU'LL PAY LATER!!"
It's just that the brokenness is *initially* hidden in the trash area
of debugging symbol memory or something.

How about just doing no optimization, but also no debugging symbols?

if it still breaks, then there is a serious code bug (upstream!) and
in my opinion, we should not release.