On 2026-03-13, Simon Josefsson wrote:
> Would it be useful to only send these e-mails when some additional
> information is available to allow further debugging?

I suspect the script used currently is rather simplistic, so it might be
a lot of complicated work to get that extra level of functionality for
the hopefully infrequent cases of missing diffoscope output, build logs,
etc. which we would generally prefer to fix at the source if
possible. It might even lead to another level of complications which
introduce additional issues in the reporting process...


> Right now all I can see below is two build logs, but that is not
> sufficient for me to debug further or resolv this.
>
> Diffoscope output between the packages is what is normally is required.

Indeed.

If there is a problem (and not just a fluke glitch), generally one would
need to figure out how to test it locally as well, including running
diffoscope...


> I've rescheduled the reproducible build, so maybe it appears later on.

Looks like the rescheduled builds succeeded, at least!


> I don't understand when diffoscope output appears on
> tests.reproducible-builds.org, and when it doesn't.  It seems somewhat
> random to me, but there is probably some logic behind it.

There can be a large and wide variety of causes, some of which are
non-deterministic ... e.g. the current load level of the system running
diffoscope might trigger timeouts, a job that triggers out-of-disk
errors (possibly indirectly breaking other diffoscope jobs), a poorly
timed system reboot...

Some inputs consistently cause problems for diffoscope, but *most* of
these issues are infrequent and intermittant...


live well,
  vagrant

Attachment: signature.asc
Description: PGP signature

_______________________________________________
Reproducible-builds mailing list
[email protected]
https://alioth-lists.debian.net/cgi-bin/mailman/listinfo/reproducible-builds

Reply via email to