Latent bugs in armel, armhf packages built before t64 transition

[Chris Hofstaedtler]

Hi,

while investigating a test failure in ksh93u+m, it became clear that
packages last built before the time_t-64bit transition can have
latent bugs.
They might very well now FTBFS or fail at runtime (autopkgtest time
or later).

I believe FTBFS are bugs are probably caught by periodic rebuilds
done by Lucas, Santiago et al.

However I am not aware of any autopkgtests ran against these
rebuilds.

In the case of ksh93u+m, the tests pass in version 1.0.8-1 (built
before t64 transition), but fail once that unchanged source is
rebuilt.

I would suggest to rebuild _all_ binary packages last built before
the t64 transition to catch such problems, before releasing trixie.


I will not pursue this any further but hope the ARM porters can pick
up this topic.

Chris

[Emanuele Rocca]

[ Martin added to CC ]

Post by Chris Hofstaedtler
while investigating a test failure in ksh93u+m, it became clear that
packages last built before the time_t-64bit transition can have
latent bugs.
They might very well now FTBFS or fail at runtime (autopkgtest time
or later).
I believe FTBFS are bugs are probably caught by periodic rebuilds
done by Lucas, Santiago et al.
However I am not aware of any autopkgtests ran against these
rebuilds.
In the case of ksh93u+m, the tests pass in version 1.0.8-1 (built
before t64 transition), but fail once that unchanged source is
rebuilt.
I would suggest to rebuild _all_ binary packages last built before
the t64 transition to catch such problems, before releasing trixie.

Thanks Chris for bringing this up.

So basically what is needed is an archive rebuild and running the
autopkgtests for all packages.

Martin, I think you and your employer were looking for ways to help the
armhf/armel ports. This looks like a great one! :)

Emanuele

[Martin]

Post by Emanuele Rocca
Martin, I think you and your employer were looking for ways to help the
armhf/armel ports. This looks like a great one! :)

Noted ;-)

[Leandro Cunha]

Hi,

Post by Chris Hofstaedtler
Hi,
while investigating a test failure in ksh93u+m, it became clear that
packages last built before the time_t-64bit transition can have
latent bugs.
They might very well now FTBFS or fail at runtime (autopkgtest time
or later).
I believe FTBFS are bugs are probably caught by periodic rebuilds
done by Lucas, Santiago et al.
However I am not aware of any autopkgtests ran against these
rebuilds.
In the case of ksh93u+m, the tests pass in version 1.0.8-1 (built
before t64 transition), but fail once that unchanged source is
rebuilt.
I would suggest to rebuild _all_ binary packages last built before
the t64 transition to catch such problems, before releasing trixie.
I will not pursue this any further but hope the ARM porters can pick
up this topic.
Chris

I've already fixed a package via NMU and was looking at some links. I
have a filter that lists the packages that have this issue to direct
the work.
https://bugs.debian.org/cgi-bin/pkgreport.cgi?users=debian-***@lists.debian.org;tag=time-t

Especially for those who follow the list and want to help.

https://wiki.debian.org/ReleaseGoals/64bit-time
https://www.thkukuk.de/blog/Y2038_glibc_utmp_64bit/