Meeting Minutes - Office Hours for CTI - 2026-08-28

[Zack Weinberg]

I'm responding only briefly to this subthread because I have very
little to add; in particular, this is the first I'm hearing of any
threats of physical violence, and I'm not going to stick my oar in
on anything that serious without knowing a lot more detail.

(Please DO NOT tell me anything about that in private.  If it cannot
be said in public, it is not appropriate to tell *me*, some random
occasional contributor, about it at all.)

On Fri, Sep 4, 2026, at 3:33 PM, Siddhesh Poyarekar wrote:
> On 2026-09-04 14:42, Andrew Pinski wrote:
>>> No it's not.  We've realized that there's no value in moving all
>>> of our eggs into one (LF IT) basket, only the ones where we want
>>> to be able to scale and have a dedicated IT team manage it.  For
>>> example, I don't see the point of ever moving buildbots (primary
>>> or workers) or bunsen over to LF IT.
...
(quoting <https://sourceware.org/pipermail/libc-alpha/2026-January/174462.html>)

> Experimental and low volume/impact uses like the forge are exactly
> what we'll continue using sourceware for; it's a great place to
> pilot infrastructure related experiments where we have root access
> and can try different things.  The LF infrastructure provides stable
> funding and scalability for core services like git, email and bugzilla.

I reiterate that to me this is exactly backwards.  You're proposing to
take the riskiest, most disruptive, and most difficult to revert step
of a hosting move as the *first* step.  How can that possibly be a
good idea?

The first step should be read-only mirrors; the second step should be
experimental, non-critical services; and at that point we should wait
at least a couple years to see what problems arise.  Even in the most
dire potential future anyone's put forward, I don't see any need to hurry.

>> This seems like you are misunderstanding my points about
>> communication.
>
> This wasn't a problem the larger community even cared about, so
> having broad discussions about this were not even something that was
> considered necessary.

It is not clear to me what problem you [Siddhesh] are saying you
thought the larger community didn't care about.  Is it the final
disposition of ancillary services, or is it the root issue, "we're
worried about sourceware not being a good place to host glibc anymore"?

Considering how contentious this has been since day one, assuming that
*anything* related to it "wasn't a problem the larger community even
cared about" seems to have been a grave error on your part.

> Since January though, we've tried to be more proactive and held
> office hours and posted minutes weekly.  I've been relying on Carlos
> doing all of the outward comms, but I think I need to step up here
> and ease his burden given the number of things he's covering at the
> moment.

Posting minutes of synchronous meetings after the fact is not good
enough.  You need to *stop holding synchronous meetings*.  You need to
start having all conversations related to this mess *exclusively* on
libc-alpha.

If there are any other mailing lists, or other asynchronous
communication channels, being used to host parts of this argument,
that's understandable -- it would be silly not to consider this
on-topic for overseers@sourceware.org, for instance -- but libc-alpha
needs to be consistently cc:ed on *every message*.

zw