Project:Quality Assurance/Meeting Summaries

This page summarizes previous meetings.

Summary of Wednesday April 23, 2014
Chair: creffett

Present: creffett, Pinkbyte, Tommy[D], TomWij, ulm, wired, zlogene

Absent: bonsaikitten, WilliamH, Zero_Chaos

Secretary: TomWij

Hacked pkgconfig files
After a recap of our thoughts and discussing how to make maintainers aware of locally modified pkgconfig files, we have voted on the following statement:

Encourage upstreaming pkgconfig files now and add a topic on -dev ML about prefixing pkgconfig files with gentoo- (eg. gentoo-${PN}.pn), which are not accepted.

6 for, 0 against, 1 abstain, 3 absent. Motion passed.

Unclarities wrt what "QA team" is
"QA team" is every member unless there is a disagreement; if the deputy or lead takes over, he takes over the term; if we vote and pass, the voters become "QA team".

7 for, 0 against, 0 abstain, 3 absent. Motion passed.

Move QA policies to the devmanual
QA team moves everything except internal policies to devmanual.

7 for, 0 against, 0 abstain, 3 absent. Motion passed.

After creffett's request for volunteers, wired has stepped up.

Documenting QA policy/workflow
Work towards creating one or two Wiki pages (perhaps a FAQ if it fits that format better). Those would then contain how to reach QA (a bug with clear information), as well as examples of what to (and what not to) contact QA with. As for whether the pages are in the right place and what on them is right or wrong, we can still work that out outside the meeting (or when there are disagreements, sort it out further at a future meeting).

7 for, 0 against, 0 abstain, 3 absent. Motion passed.

Open floor
No open floor items were brought forward.

Summary of Wednesday February 19, 2014
Present: creffett, Pinkbyte, Tommy[D], TomWij, ulm, wired, zlogene

Absent: bonsaikitten, WilliamH, Zero_Chaos

Deprecating/Banning EAPIs
Discussed which EAPIs to recommend to the council to deprecate or ban.

Banning EAPI 1: 6 for, 1 against, 0 abstain, 3 absent. Motion passed.

Banning EAPI 2: 3 for, 2 against, 2 abstain, 3 absent. Motion did not pass.

Deprecating EAPI 0 (pending discussion with toolchain): 7 for, 0 against, 0 abstain, 3 absent. Motion passed.

Deprecating EAPI 3: 4 for, 1 against, 2 abstain, 3 absent. Motion passed.

Revisit policy on dropping packages to unstable when stabilization takes too long
Postponed until we have Zero_Chaos available to articulate his objections.

vapier stabilizing on experimental arches
Postponed until the council actually meets.

Multislot issue
Forum poll resulted in the vast majority of respondents not using multislot, and discussion with toolchain resulted in dirtyepic saying that he was okay with dropping the flag. Team agreed that the eclass can keep the code, so that toolchain can use multislot in their overlay if they want, but the multislot flag should be removed from the Portage tree. 7 for, 0 against, 0 abstain, 3 missing. Motion passed.

Tracker bug for this issue: https://bugs.gentoo.org/show_bug.cgi?id=507808

GTK flag situation
Three votes were taken:

Should we require gtk to use versioned USE flags? (gtk2, gtk3, ...)? 6 for, 1 abstain, 3 absent. Motion passes.

Should we allow USE=gtk to mean "Use gtk2" due to the number of ebuilds already using it for this purpose? 5 for, 1 against, 1 abstain. Motion passes.

Should this be a general policy? i.e. in cases where something can link against one or more versions of a library, there should be a versioned USE flag? 4 for, 2 against, 1 abstain, 3 missing. Motion passes.

The last choice was supplemented by a fourth vote, which changed the wording to recommending the use of versioned USE flags and requiring that it be discussed with QA. 5 for, 1 against, 1 abstain, 3 missing. Motion passes.

Open floor
creffett reprimanded the team for complaining about each other to him.

Summary of Wednesday January 29, 2014
Present: creffett, Pinkbyte, Tommy[D], TomWij, ulm, WilliamH, wired, zlogene

Absent: bonsaikitten, Zero_Chaos

Log: gentoo-qa-team-meeting-log-20140129.txt

The lead creffett appointed TomWij as deputy lead
When creffett is out (which will be a lot of weekends this semester) and a decision has to be made, TomWij is in charge; and anything TomWij does as deputy lead carries the same weight as creffett's decisions.

Meeting scheduling (Third Wednesday, 1900 UTC)
Third Wednesday of the month (which is a week after Council's meeting) at 1900 UTC.

8 QA members agreed, 0 rejected, 0 abstains, 0 didn't vote; 2 absent. Motion passed.

Policymaking workflow (by creffett and Tommy[D])
When a person brings us problem, we look into the problem and discuss it at meeting. If there is no policy on problem, we make policy; if the policy or documentation is unclear, we update it. If the policy is actively being ignored we politely ask the person to stop. It is more reactive than proactive, this does not preclude emergency action on our part. This will give the team the time to work out basic rules and workflows, we might do more proactive tasks later one, if there is any need for them. However, if we think somebody is breaking the tree, we can ask them to stop and/or undo what they did pending a review at a QA meeting.

6 QA members agreed, 0 rejected, 0 abstains, 2 didn't vote; 2 absent. Motion passed.

Amount of communication expected from the QA team when making changes to maintainer's packages
We fix and send a friendly reminder for trivial fixes; open bug, wait 2 weeks, make a change for larger but non-critical fixes; we make a change and send a notification for critical fixes.

8 QA members agreed, 0 rejected, 0 abstains, 0 didn't vote; 2 absent. Motion passed.

GTK USE flag situation (gtk, gtk2, gtk3; relevant to )
We tentatively recommend "gtk means gtk2, gtk3 means gtk3, burn gtk2 flag, transition flags as appropriate" and discussing this further on @-dev.

Voting on this is unclear, however we have a majority that had no objections and wants to see discussion on the mailing list; there is no motion to pass here, it is merely a recommendation.

Official recommendation for how to recommend optional RDEPs (relevant to )
This yields a bikeshed between (not limiting to, in no particular order) README.gentoo, elog and optfeature as to which approach to use for this; and thus, for the moment, we agree that using USE-deps for optional packages is bad. Except in certain case-by-case circumstances, where an acknowledgement by the QA team is needed (which is then documented in a comment in the ebuild).

8 QA members agreed, 0 rejected, 0 abstains, 0 didn't vote; 2 absent. Motion passed.

How do we want to communicate policy/policy changes made during meetings?
gentoo-dev-announce, gentoo-dev, Gentoo Wiki, maybe a blog post, ...

No voting really happened, it comes down to communicate them on the required as well as on multiple appropriate channels such that we reach the right people.

What large projects do we want to tackle as a team?
Due to the length of the discussion, we consider "Possible Future Project" in Project:Quality_Assurance/Current_projects as sufficient to list the large projects; no actual discussion is crucially needed here.

Stabilization thread (TL;DR: Some slow arch team's queues are growing, how do we deal with that?)
Due to the multiple options / topics involved the early discussion mixed some of the topics; where the general idea is that we need get more people interested in joining arch teams (advertisement, PR, ...) as well as make sure that the arch team's work is more fit to the situation (more work on important, less work on non-important, ...) and that the naming of ~ can be confusing ("unstable" vs "testing" vs "current" vs ...).

The later part of the discussion however came up with a statement by Tommy[D] that we vote on: "Allow the maintainer to drop stable keyword or last stable version, if arch team does not respond within 90 days; if it breaks the dependency tree, then the maintainer has to work with maintainers of depending packages before dropping keyword/last stable version."

7 QA members agreed, 0 rejected, 1 abstains, 0 didn't vote; 2 absent. Motion passed.

Multislot issue
We are concerned about the "multislot" USE breaking metadata in-tree, and we would like to remove the use of multislot to a toolchain overlay pending proper dynamic-SLOT support in an EAPI. However, we need more information on this before we can make a proper recommendation. ulm will create a forum poll asking users whether the removal of multislot would break their workflow, and Pinkbyte will email toolchain to ask if they anticipate the removal of multislot causing any problems. Further discussion postponed until our next meeting.

Stabling on non-stable arches
Concerns were raised over vapier stabling packages on arches marked as non-stable. We will wait for the council to vote on this (https://bugs.gentoo.org/show_bug.cgi?id=498332) before we take any action.

Meeting lengths
We will, in the future, keep meetings under 2 hours, unless the team votes to continue.

EAPI deprecation/banning
Put off until next meeting.