On Thu, Dec 12, 2013 at 12:04:35PM +0100, Mark Trompell wrote:
> How much time would you think will it keep to get on track? I think as
> you already played with fedora imports and centos/redhat works too,
> Fedora won't be a big issue. Though systemd and especially the
> usr-merge that followed its introducion will need some extra love.
> I think after the first import, regular updates and bumping fedora
> versions shouldn't be that hard (mainly because people will know
> mirrorball by then)

So, "get on track" is where the unknown work is.  We don't know how
much work it will take to adapt to all the recent Fedora changes.
Dick has two proposed patches that he's pointed to (thanks) that we
need to look at, though I don't know whether the test suite works
with them.  This part is where I want to propose joint work between
Foresight developers and SAS developers.

After that, updates are usually not a big problem.  Just run a robot
to run mirrorball, and occasionally it fails and imports stop until
someone figures out why and fixes it.  Sometimes the required changes
are to the factories that drive the import process; more often, the
safety checks in mirrorball have flagged a possible inconsistency
that has to be investigated and either reported upstream as a bug,
is temporary and we should just wait for it to resolve (e.g. packages
released out of sync with sources), or a real change that requires
an exception to be added to mirrorball config.  This is not, normally,
a daily process.  With CentOS, it isn't even every week; most of the
work is identifying new exceptions due to packaging changes at point
releases.  But since Fedora allows more change, I see that this could
happen slightly more often.  This is where I'd expect that Foresight
developers would do the normal diagnosis, with help figuring out
what to do in exceptional cases.

> I think I can dedicate some time a day, but unfortunatly all sas guys
> are still sleeping at that time.

Working further around the clock isn't a bad thing.  I imagine the
robot sending email to a team when things break.  If you get an email,
you could investigate, and even if you haven't fixed it, you could
summarize what you learned in email to this list and the next person
could pick up from there.

This is where our too-IRC-driven culture has gotten in the way.  Not
enough use of the mailing lists.  It has not only made it hard for
those not always on IRC to follow what's happening, but has made us
look like we don't exist, because people look at the mailing list
and say that we're dead, because on the mailing list it looks that
way.

_______________________________________________
Foresight-devel mailing list
[email protected]
https://lists.foresightlinux.org/mailman/listinfo/foresight-devel

Reply via email to