[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: #932795 - Setting up minimum build system expectations?

Hello all,

Dear Release Team members, I am reaching out to see what is your
position regarding bug #932795, which in turn references
#907829. Thing is, I want to push a bit before we (Tech-Ctte) to have
a possible decision for our next meeting (2019.08.21).

In short - This bug started off quite aggressively between two
developers, and I don't think it's easy to deescalate. But I think
that there is a good way to solve the situation - _if_ the Release
Team is willing to accept one more power / responsability.

I am expressly _not_ including 932795@bugs.debian.org as a recipient
to this mail, as I'd like not to resonate too much with the affected
people at first. Of course, I am including an open mailing list. /me
crosses fingers ;-)

If, in a fashion similar to issuing release qualifications for
architectures, the Release Team indicates which is the minimum
acceptable build environment for a FTBFS to be considered RC.

I don't want to repeat all the reasoning that was given on the
{bug,list}, so I'll just point at a couple of meaningful messages:

- I think Ian Jackson's mail
  <23862.65026.843324.814828@chiark.greenend.org.uk> is quite sensible
  and sane. Of course, it would be up to the RT to define the minimum
  levels considered "reasonable" and "supported" for each
  release. Of course, this could include RAM, number of cores, free
  disk space available — And be flexible to add other criteria not yet
  considered if deemed necessary

- We might at some point hit packages that FTBFS with 4GB RAM, which
  would be tricky -to say the least- to build them in 32-bit systems,
  as Ansgar mentioned in <87ftmwd1yh.fsf@43-1.org>

- In this message, the submitter explains what is requested from the
  TC. I am, however, reaching out to the RT because I think this is a
  problem that should be addressed, and not on a case-by-case
  basis. <20190725112242.og3gey4fmpwnt3od@nucold>

- The submitter makes a point on the amount of packages presenting
  build failure (4%), plus the economic difference it takes for him to
  hire a better server. https://people.debian.org/~sanvila/single-cpu/

Thank you all for your comments,

- Gunnar, wearing his tech-ctte hat

Attachment: signature.asc
Description: PGP signature

Reply to: