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

[Zack Weinberg]

On Sun, Sep 27, 2026, at 10:21 PM, Siddhesh Poyarekar wrote:
> On 2026-09-27 21:28, Zack Weinberg wrote:
>> The decision process is NOT done.  You, as the proponent of the
>> proposal, *cannot* unilaterally declare consensus reached while
>> there are sustained objections; that's a conflict of interest.
>> Consensus does
>
> I'm not declaring consensus, I'm stating that the present sustained
> objections are not negotiable and [...]

Negotiable or not, sustained objections mean you may not proceed to
implementation.

I really have to insist on this point: Neither you, nor any of the other
advocates for the move, are entitled to decide to proceed despite a
sustained objection from any member of the community, no matter how
intractable you think the disagreement is, nor how unreasonable the
objection seems to you, nor for any other reason. This principle is more
important than any outcome.

> The fact that the overwhelming majority of the current contributors
> are onboard the decision,

is irrelevant.

>> You do not get to make that call.  You are (one of) the proponents of
>> the proposal on the table; you cannot reject a sustained objection to
>> your own proposal.  That is a blatant conflict of interest.
>
> I'm not making that call (I know I don't have unilateral authority to
> do that), what I'm saying is that I don't consider that as a reason to
> change my mind about it.

Tomayto, tomahto.  You keep telling people a decision has been made and
the move can proceed.  I'm going to keep saying you can't make that call
until you stop trying to make that call.

>> Also, it's *your* order of operations that doesn't make any sense!
>> How can it possibly be sensible to move the essential services
>> first??! It's *horrifying* to me that nobody brought this up before
>> now. It suggests nobody on your side of the argument actually thought
>> through the risks of the migration itself.
>
> I consider it a higher risk position to have critical services being
> maintained by part time volunteer administrators.

The project's critical services have been maintained acceptably well by
part time volunteer administrators for something like 25 years now. I
think they can be trusted to do it for at least a couple more years--
long enough for you to test out the process of the move itself, and give
the new infra a thorough shakedown.

Rushing to execute a process that hasn't been thoroughly tested on non-
critical services is asking for a disaster on the level of "oops, there
was a typo in the migration script that wiped the primary Git repo on
*both* the old machine and the new one, and it turns out the backups
haven't been working for years either, and now we have to try to
reconstruct things by comparing all the clones on contributors'
machines."  Is that really a risk you're willing to run?

> Like before, I'm not saying this from any position of authority since
> I don't have one.  I am contesting your opinion of "there's no policy,
> so there can be no decision without a consensus".  The consensus
> process for glibc is for glibc development, not governance or
> infrastructure discussions. The GNU maintainers are responsible for
> that and we've had a split vote there (IIRC), further to which pretty
> much all of the active developers have voiced their support in favour
> of the decision.

I am actually saying something slightly different: I'm saying that *this
particular decision* is so consequential and so controversial that only
consensus of the entire community is an acceptable basis to move forward
with it.  The GNU maintainers by themselves are not enough, a majority -
even a supermajority - of the active developers is not enough.

And I'm also saying that *telling me my objection is invalid* is not an
okay thing for you to do, because you're conflicted.  You can't
simultaneously be an advocate of a proposal and an arbiter of whether
objections to the proposal are invalid.

>> I am quite confident in saying is that a decision of this magnitude
>> is *unprecedented* in at least the past 15 years, probably longer --
>> how long ago was it Ulrich Drepper quit being sole proprietor? -- and
>> that we do not have any previously agreed upon framework for major
>> decision making.
>
> You're again conflating governance and administration with
> development.

Governance and administration decisions are *more important* than
development decisions, and therefore it is more important that they be
carried out in public, and be both fair and seen to be fair.

> No, the LF does not spend money for the projects, it's member
> companies that sign up to sponsor projects.  Each project (so that
> would be us in the CTI) makes a case for funding with the member
> companies, telling them where their money will be spent.  Critical
> services have a direct line to the businesses of these member
> companies, which makes it worthwhile for them to invest in the
> project.  Non-critical services don't quite have the same importance
> to their businesses; replacing the wiki for example with a git-managed
> sphinx instance for example is not interesting enough for a business.

So you go back to the company or companies that wants to fund the
critical services and you say "okay, this sounds good on paper, but we
want to test out the infrastructure you're offering us for a good long
while before we start using it for anything critical", and if they say
"no, we won't let you do that", then they're trying to sell you a pig in
a poke and you should walk away from the deal.

Is this really such an alien notion to you?

>> Are you sure this isn't the sunk cost fallacy talking?  You _have_
>> spent five years trying to make this work ... and maybe that means
>> it's time to start seriously considering alternatives.
>
> It's not.  Even if I step away from this all and think about my
> opinion as a glibc contributor, I think LF IT infrastructure is a
> vastly superior solution for all GNU toolchain projects.

But you've spent five years trying to persuade people that LF IT is the
best option, and right now it looks like you're not going to be able
to persuade the entire community, and that means the move's not happening.
So maybe you should start putting some thought into what the next best
option might be.

zw