CTI - Making a decision for glibc.

[Arsen Arsenović]

-text follows this line--
<#secure method=pgpmime mode=sign>
Hi,

Alexandre Oliva <oliva@gnu.org> writes:

> On Jan 28, 2026, DJ Delorie <dj@redhat.com> wrote:
>
>> Sourceware is also SaaSS
>
> I believe that's correct, overall.  It is concerning indeed.

I see no valid concern in the Sourceware context (and, even by the looks
of it, in the CTI/LF context.  I'm not a fan of the LF, and I'd prefer
personally if nothing moved to their infrastructure, but Andreas is
right in that my opinion on the matter doesn't actually count as I'm an
infrequent contributor, and LF seems to be okay with respecting the
requirements we do have).

I, frankly, support offloading (substituting, if you will) work of
maintainers onto (with) Sourceware infrastructure.

The infrastructure need be Free, of course, and the processes
reproducible (for practical and principal reasons), but it is not "good"
to require drudgery from contributors.  And it is.

We only lose by forcing maintainers to perform work that could be done
automatically.

> AFAIK we (glibc) are not using any of its SaaSSs.
>
>> so we should immediately move off it
>
> If we are using any of its SaaSSs, discontinuing their use is probably
> much saner than rushing to move out, especially to a provider that
> would make that worse.
>
>> running gcc-as-a-service on my web site
>
> Yeah, that sounds like SaaSS to me.

It sounds practically insignificant to me.  I have to say that I'd have
rejected the request.

Compiler Explorer is an example of such a service.  It is immensely
useful to us compiler hackers as it allows sharing and comparing with a
variety of compilers.

And, it is not practical to replace a compiler with Compiler Explorer,
so if one of the ess-es in SaaSS is "substitute", then CE cannot qualify
for it.

CE is, of course, free software.

>> It has nothing to do with "propriatary" and everything to do with
>> whether someone else's computer is running the software you use.
>
> It's not so much about whose computer it is, but about who has control
> over it.
>
> Using a rented computer from a VPS is not SaaSS, if you control it.
>
> Hiring services controlled by others to do your own computing is SaaSS.
>
> Publishing a GIT repository (publishing) or hosting a mailing list
> (communication and publishing) are not SaaSS.
>
> But performing server-side GIT merges, for example, is SaaSS if the
> server is operated by a third party.

... and here we have a perfect example.  Merging patches requires
tonnes of drudgery, specifically the drudgery of testing (and sometimes
testing twice and comparing as for GCC for instance unfortunately) on a
variety of machines.

Pushing this to contributors means no contributors can ever actually
contribute, as very few have access to the cfarm, and getting cfarm
access is too high a barrier to entry, or that they simply don't test
extensively (as happens in practice).

Pushing this onto maintainers is frankly a waste of scarce time of
already overburdened people.

Pushing this drudgery, these steps that can be performed perfectly
automatically, onto a machine, is exactly the right thing to do.  This
reduces maintainer load and *INCREASES* quality, as the chance of error
is significantly lowered.  As we expect contributors to be, well,
people, in different circumstances, these machines logically fall into
the hands of their common factor: the project being contributed to.

Obviously, this doesn't mean that contributors shouldn't or couldn't run
tests.  That'd be ridiculous.  But, as a real-world example, I frankly
don't want to have to deal with, say, AIX, while developing a patch.
Issues with my patches on AIX et al. can be discovered by an automated
system without wasting my time in the happy path.  The same tools that
such an automation (commonly called CI) would use are already present in
the GCC repository and would be invokable by me on the cfarm AIX machine
(assuming it's still around; I haven't checked).

If the above is not SaaSS, then no SaaSS was proposed.  If it is, then
it's not a useful category, as it's in clear conflict with something
actually useful.

Have a lovely night.
-- 
Arsen Arsenović