Created attachment 376806 [details, diff] [PATCH] ebuild-maintenance: Add developer stabilization exception for AMD64/X86 arches. As discussed in the recent gentoo-project ML and in #gentoo-qa, yesterday; let us document this exception properly, to avoid us from forgetting it over time... Attached you can find a suggestion for a patch. Thank you in advance.
I presume this is acked by QA? (we really need to find a better way for devmanual team to coordinate with QA. How about having the lead or 2 members of QA sign-off the patches so i stop annoying you everytime?)
This needs to be acked primarily by arch teams, I think. The wording could also be a bit clearer: It first talks about "open bugs for stabilization" and then continues with "keywording of your packages". Presumably, both keywording and stabilisation should be allowed?
(In reply to Ulrich Müller from comment #2) > This needs to be acked primarily by arch teams, I think. > > The wording could also be a bit clearer: It first talks about "open bugs for > stabilization" and then continues with "keywording of your packages". > Presumably, both keywording and stabilisation should be allowed? from my experience in amd64, the rule is that you are allowed to stabilize the package yourself, if you have the hardware and a stable environment to test it. For keywording (aka ~testing) as long as you have the hardware, you can mark it ~amd64 yourself.
Another (less important) change; added here as a comment, to not forget it: @ulm | kingtaco's e-mail was from 2007-11-14 (UTC) or 2007-11-13 (his local time) @ulm | and the IRC conversation was at 2007-10-10 Yes, we should look into the wording and afterwards get an ack from the arch teams; maybe, after that is done, we could have QA ack it too in the QA meeting. I filed this bug for tracking purposes, don't want to make a commit without ack; but indeed, who an ack is needed from should be worked out further...
Arch teams, QA, ...: Do you acknowledge the patch and/or do you have any remarks?
GLEP 40 needs updating.
Find below my opinion about this, that I mailed back when pacho asked about this on the x86 mail list: ---- As member of the x86 team I'm personally not fond of letting devs stabilize their own packages (although I think there is/was one herd that is allowed for x86 to do this). Too often to be comfortable I find things broken that "works for the dev" (read: their testing procedures), or do stupid things. Arch testing is for QA, letting devs judge their own work (see my comment above) is blowing this QA to pieces and you could as well disband all the arch testing teams. I run a couple of systems as stable for a reason, and QA is an important part of this. These days many packages have broken test suites (slowing arch testing down considerably for me) and shows that many devs don't take QA too serious. And now they want to do the last bit of QA themselves? My opinion is that devs have to ask the arch team if they want to stabilize the package themselves, but also need a very good reason(!) for this. This also needs to be communicated back into the team, discussed and documented somewhere. ---- In support of this opinion, check bug #513308 (the glibc/gentoo-sources "QA" mess that was generated last week).
Created attachment 393472 [details, diff] 0001-Update-arch-stabilization-exceptions-on-the-Keywordi.patch
Created attachment 393474 [details, diff] 0002-Add-a-Stabilization-section-to-the-AMD64-arch-page-b.patch
Created attachment 393476 [details, diff] 0003-Add-a-Stabilization-section-to-the-x86-arch-page-bug.patch
I found the bug after I wrote the patch, but that's OK: we patched different things. Even if we don't want to make this change, I would really like to know one way or another. Arch teams, if you don't want us doing this, just remove yourself from the description and say so.
It's been a while. Arch teams didn't ack but according to http://www.gossamer-threads.com/lists/gentoo/dev/296420?page=last this is an acceptable policy Thus I will merge these patches.
This "you failed to respond in three minutes so you must agree" is not very productive. I would prefer some form of code review before packages are stabilised. If the arch team isn't responsive and you can get another dev to look over your work, fine. Just doing it yourself leads to tragedy.
Err, isn't this topic flying around for months? :/
+<title>Stabilization</title> +<body> + Usually only the arch teams may mark a package stable (see <uri + link="::keywording#Moving from ~arch to arch" />). The AMD64 team + allows maintainers to stabilize their own packages as long as they + have the appropriate hardware and a stable system with which to + test. +</body> +</section> + +<section> And as far as I remember this is ok for amd64
(In reply to Pacho Ramos from comment #15) > +<title>Stabilization</title> > +<body> > + Usually only the arch teams may mark a package stable (see <uri > + link="::keywording#Moving from ~arch to arch" />). The AMD64 team > + allows maintainers to stabilize their own packages as long as they > + have the appropriate hardware and a stable system with which to > + test. > +</body> > +</section> > + > +<section> > > And as far as I remember this is ok for amd64 it should. i don't recall the leader revoking that.
For x86 I would like to see a note regarding processor features. Now after doing x86 arch work for 1 year I found several problems in packages assuming SSE2 support. And seeing devs marking a package stable for x86 because they tried in a VM on their current hardware which is SSE2-capable doesn't help.
(In reply to Michael Orlitzky from comment #8) > Created attachment 393472 [details, diff] [details, diff] > 0001-Update-arch-stabilization-exceptions-on-the-Keywordi.patch $ curl -Ls https://bugs.gentoo.org/attachment.cgi?id=393472 | git am -sS Applying: Update arch stabilization exceptions on the "Keywording" page (bug #510198). error: patch failed: keywording/text.xml:222 error: keywording/text.xml: patch does not apply Patch failed at 0001 Update arch stabilization exceptions on the "Keywording" page (bug #510198). [...] Please rebase onto master. Also, please follow the style guide, and make sure that it validates, and add a signoff.
(In reply to Ulrich Müller from comment #18) > > $ curl -Ls https://bugs.gentoo.org/attachment.cgi?id=393472 | git am -sS > Applying: Update arch stabilization exceptions on the "Keywording" page (bug > #510198). > error: patch failed: keywording/text.xml:222 > error: keywording/text.xml: patch does not apply > Patch failed at 0001 Update arch stabilization exceptions on the > "Keywording" page (bug #510198). > [...] > > Please rebase onto master. > > Also, please follow the style guide, and make sure that it validates, and > add a signoff. You might have better luck if you didn't wait five years to apply patches. Since you've already opened your own conflicting PR on github making these same changes, I don't think I'll bother.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/proj/devmanual.git/commit/?id=b696fdc696574cd1fec19ff572c951c464fe10be commit b696fdc696574cd1fec19ff572c951c464fe10be Author: Tom Wijsman <TomWij@gentoo.org> AuthorDate: 2014-05-13 00:52:15 +0000 Commit: Ulrich Müller <ulm@gentoo.org> CommitDate: 2020-01-25 05:53:46 +0000 keywording: Add developer stabilization exception for AMD64/X86 arches. References: - "AMD64 keywording" thread by kingtaco on the gentoo-core mailing list (2007) - app-emacs stabilization talk by kingtaco and ulm on IRC #gentoo-dev (2007) - Gentoo Council agenda item: "amd64 arch team and big bug list" (2008/03/13) Closes: https://bugs.gentoo.org/510198 Signed-off-by: Ulrich Müller <ulm@gentoo.org> keywording/text.xml | 7 +++++++ 1 file changed, 7 insertions(+)