Working together and gaining trust

[Carlos O'Donell]

On 3/6/26 8:17 AM, Mark Wielaard via Overseers wrote:
> Sorry for not immediately replying to this discussion. I had a death
> in the family which took a lot of (emotional) energy. And then there
> was of course the Sourceware hardware refresh cycle to finalize last
> week.

I'm sorry to hear that :-(

Thank you for your support, but please take the time you need for your
family.
  
> But I am happy the emotions seem to have calmed down and the proposal
> now technically just amounts to a glibc git branch cleanup, which seem
> way less controversial. But these controversial proposals seem to come
> back every 9 to 15 months out of nowhere. So we probably should
> discuss how to resolve this properly without all the drama. Also
> because it eats up lots of energy which distracts from the actual
> infrastructure improvements. So I hope the FSF, SFC, OpenSSF and LF
> can just sit down and discuss how to efficiently move forward.

Please allow me to me clarify.

The branch cleanup is the first step to get a gitolite configuration
that is valid for a transition to CTI provided services.

Branch cleanup is not an end to itself, but a start.

We are proceeding down the path outlined in the announcement email.

> So the discussion is completely different from five years ago.
> Ironically Sourceware became so "professional" that some people are
> now apparently seeing it as "the other" that might control their
> compute. But in reality the Sourceware PLC, overseers and admins are
> still just members of the various hosted projects working on our
> infrastructure together.

Even if Sourceware is different from where it was five years ago, and
that's good, it still does not meet the requirements for security,
robustness, isolation and sustainability that the GNU Toolchain or
glibc need.

Are there services where we can work together? Absolutely, and I think
it's critical that Sourceware continue for those projects that decide
to use Sourceware and for other services where we need hands on
development efforts e.g. new workflows, pre/post-commit CI.

Are there core services which are going to be expensive to fund
sustainably and need a dedicated IT team to handle the isolation
and maintenance?

Yes, and CTI exists to meet those core service requirements with a
different and sustainable funding model.

> On Tue, Feb 17, 2026 at 07:58:37AM -0500, Siddhesh Poyarekar wrote:
>> On 2026-02-17 06:36, Mark Wielaard wrote:
>>> You make it sound like we didn't like the idea of making Sourceware
>>> independent from Red Hat or that reaching out to the SFC to setup a
>>> proper Sourceware organization was somehow not part of this plan. I
>>> believe all overseers (whether they happened to work for Red Hat or
>>> not) were initially enthusiastic about the idea to get more resources
>>> for Sourceware. We even started making public resource estimates and
>>> technical plans [1] [2]. The only detail some of us were worried about
>>
>> I don't deny that you didn't dislike the idea of being solely
>> dependent on Red Hat for funding but again, you're sharing links
>> that simply reinforce my conviction that reaching out to SFC was a
>> reaction to disliking the LF IT proposal, which you were made aware
>> of much before you let on in these public conversations.  It's
>> absolutely OK if it was a reaction, it's fine too if you may have
>> had reached out to SFC before or around the time that you got to
>> know about the LF IT proposal.  But pretending that you didn't know
>> about it until much later is disingenuous and doesn't invoke trust.
> 
> This is 5 years ago, so we might misremember the timeline or what was
> known to who. All I can say is that I don't remember involving the LF
> IT really had been part of any of the discussions. And even if I knew
> I wouldn't be against that because I know Konstantine and would have
> liked the resources to hire him and working together. We did discuss
> setting up Sourceware like kernel.org which also is a non-profit
> getting sponsorship money through the LF. Which is why we reached out
> to the SFC to help us with that. And it was actually Carlos that
> suggested we do that so the community could hold assets and enter into
> contracts, etc.

You were aware early that I was proposing delegating cores services to
a paid IT team, and that I have always been recommending LF IT from the
start given their FOSS alignment.

I advised you at the time to seek out additional sources of funding.

You did that by reaching out to the SFC.

> Same with the public resource estimates, technical plans and
> presenting about the infrastructure services at Cauldron. We wanted to
> let potential sponsors know what the community needed and were working
> on. We just wanted to make sure that we had the organization and
> community setup to start working with more corporate sponsors if they
> would show up. Which I thought the LF plan was all about. So it
> certainly wasn't against that. The idea was for us to work together
> not against.

I agree we should work together.

What does that look like to you going forward?

>>>>> Which is why it is so important that before this CTI plan proceeds the
>>>>> FSF as the original fund raising organization for the GNU Toolchain
>>>>> signs off on it and makes sure that a clear charter and board that
>>>>> works in the interest of the community and guarantees the money is
>>>>> spend on Free Software.
>>>>
>>>> Nope, you're just inventing new governance rules now and creating
>>>> friction where none exists.
>>>
>>> I am not inventing, the friction is already there. The FSF has real
>>> issues with this idea [3]. And given that the FSF is legally the
>>> steward of the GNU Toolchain it is only professional for the LF to
>>> discuss with the FSF if they want to take over some of that to make
>>> sure they align on the goals.
>>
>> I'm referring to your claim of needing a signoff from the FSF; I
>> don't think we need any such signoff since we satisfied their demand
>> of not calling it GNU Toolchain Infrastructure.
> 
> I don't know if just changing the name suddenly makes it legitimate. I
> guess the FSF and the LF will have to have a talk about the conditions
> that would make it so.

You have twice called into question the legitimacy of the GNU Project
to make decisions on both this mailing list and on the GCC mailing
list and both times you've been told you are not correct.

For reference:
https://inbox.sourceware.org/gcc/20250616165940.GS30295@gate.crashing.org/
https://inbox.sourceware.org/libc-alpha/c9ae9bf3-5f18-4493-8367-d613818ec916@redhat.com/

In this case you are also incorrect.

> But my main point really is that as a project we shouldn't just ignore
> the FSF or do things they clearly don't agree with. We have a good
> working relation now with the FSF and the FSF tech team, they have
> paid staff which provides various services for some of our hosted
> projects. Lets work together with them. If they point out issues that
> concern them then lets make sure we address them together. We aren't
> adversaries, we all just want to advance Free Software.

I agree we all want to advance Free Software.

We started conversations with the FSF very early to understand the
foundations position on this topic.

I'll continue to engage with the FSF and the GNU Project.
  
>> I think we've addressed all concerns other than the "I don't like
>> the LF", which I don't think we can address.  The best we can do is
>> invoke our personal currency in the community as individuals in the
>> TAC and say that we will ensure that if we're made aware of anything
>> that goes against the principles of our community (which implicitly
>> includes Free Software principles) we will do whatever it takes to
>> remedy that, including moving infrastructure out if needed.
> 
> The best you could do is fix the charter and board requirements. It is
> the shaky governance that people don't like. The charter doesn't even
> talk about Free Software. The board just seems to be about who pays
> most and then even decides who may "advise" them. Why not add some
> guardrails to the charter by describing what raised money may be used
> for. Make sure the board has at least the FSF as a member. It doesn't
> have to rely on just "personal currency", you can just put down things
> in writing you agree on.

In general "people" trust the GNU Toolchain leadership to negotiate,
like we have, for the last 30 years, to have sponsors that are aligned
with the mission and vision of the GNU Project and support the use of
FOSS.

Is the CTI charter the biggest concern you have today?

> But also the LF and OpenSSF could just work directly with the FSF,
> SFC, Sourceware and the project community so we can improve the
> infrastructure together.
This is a false dichotomy.

We can move the GNU Toolchain to CTI services using paid IT for
core infrastructure *in addition* to supporting Sourceware.

We are working with the project community.

We are improving infrastructure together.

-- 
Cheers,
Carlos.