FESCo-2008-03-27

--- Topic for #fedora-meeting is Channel is used by various Fedora groups and committees for their regular meetings | Note that meetings often get logged | For questions about using Fedora please ask in #fedora | See http://fedoraproject.org/wiki/Communicate/FedoraMeetingChannel for meeting schedule
Topic for #fedora-meeting set by quaid at Wed Mar 26 16:16:38 2008
--- bpepple has changed the topic to: FESCo meeting -- Meeting rules at http://www.fedoraproject.org/wiki/Development/Schedule/MeetingGuidelines -- Init process
* jeremy is here, though partially distracted
bpeppleFESCo meeting ping -- bpepple, caillon, c4chris, dgilmore, dwmw2, f13, jeremy, jwb, notting, spot, nirik, tibbs, warren
Hi everybody; who's around?
* tibbs here
nirik is here.
jeremy is here, though partially distracted
spot is here
jwb
* bpepple waits another minute or so before starting the meeting.
* warren here
bpeppleok, we probably get started...
--- bpepple has changed the topic to: FESCo meeting -- Any objection to this week's report from FPC at https://www.redhat.com/archives/fedora-devel-list/2008-March/msg02169.html
* dgilmore is here
bpeppleFirst up, it looks like the FPC approved 3 proposals.
tibbsI'm sure everyone saw the ascii naming flamewar.
nottingis there a simple diff of the new perl vs old perl guidelines?
tibbshttp://fedoraproject.org/wiki/PackagingDrafts/Perl?action=diff&rev2=21&rev1=1
* dwmw2 arrives
tibbsI think they started with a copy of the existing guidelines.
bpeppleAre there any objects to the FPC proposal for ASCII Naming Guidelines?
--- bpepple has changed the topic to: FPC Guidelines - http://fedoraproject.org/wiki/PackagingDrafts/ASCIINaming
nirikdoes that mean someone will need to go thru and poke all the perl packages to conform?
tibbsI don't think there's a general failure to conform.
nottingbpepple: do we have any packages that do not conform?
bpepplenotting: I'm not aware of any.
tibbsNo packages fail to conform to the ascii naming guidelines that I could see.
There is one package pending review.
niriksorry, I was meaning perl.
* warren is sad that he cannot call an RPM named 窓際族
nirik is fine with all the FPC proposals.
tibbsI do not believe there are a significant number of packages which fail to conform to the new Perl guidelines; they only codify existing practise in a more organized way than the previous guidelines.
* bpepple is fine also with the ascii proposal.
jwbi have no objections
dwmw2I think the ASCII restriction is silly.
UTF-8 ftw
warrenconfusing to ask about all three at once, perhaps easier to ask for objections?
dgilmorewarren: kinda hard for me to type
dwmw2: indeed
caillondwmw2, agreed, actually.
warrendgilmore, our deficient educational systems are blame.
dwmw2this is the 21st century, for crying out loud. Haven't the Luddites given up on ASCII yet?
jwbwarren, hardly
warrenjwb, I'm joking. =)
tibbsI will note that the FPC voted unanimously for the ascii naming guidelines.
dwmw2We should approve écolier-fonts asap
jwbwarren, ok.  too close to home for me
dwmw2tibbs: I blame mass hysteria.
tibbsI blame careful consideration.
jwbsummarize?
tibbsI think fesco should call a vote on the issue.
bpeppletibbs: agreed.
jwbtibbs, can you summarize the rationale?
dwmw2was there any good reason given for sticking with ASCII?
bpeppleCould I get a show of hands for the ascii proposal?
tibbs+1 ascii naming.
dwmw2other than "We might have bugs somewhere and we can't be bothered to find or fix them"
warrenI think it is a good pragmatic decision to allow only ASCII for now.  There are too numerous problems supporting anything more at the moment.  We might leave the door open for reconsideration in the future so not to completely discourage people from trying to come up with a workable plan.
tibbsI will not attempt to summarize the rationale of all of the committee members.
bpepple+1 ascii naming.
* jwb sighs
tibbsIf FESCo wishes each of the FPC to summarize their rationale, it can ask.
dwmw2-1. As soon as F9 is released, put écolier-fonts in rawhide and we have a whole release cycle to see if anything actually does break
jwb0
dwmw2Fedora is supposed to be about new stuff, not Luddism.
nirik+1 ascii naming (for now, guidelines can be changed later if need be)
dgilmore-1 I really think we should be ok with UTF-8
warrenI'm willing to vote +1 on ascii naming *IF* we allow écolier-fonts into rawhide afterward only as an experiment.
* rdieter thinks the folks wanting UTF-8 should then make a write up counter proposal for the FPC to consider.
dwmw2warren: what's the benefit in voting for the broken proposal only to violate it immediately afterwards? :)
jwbrdieter, that's a good idea
warrendwmw2, your idea =)
caillon-1. having specfiles that are UTF-8 (rpmlint complains if not) and then restricting a piece of it to be ASCII seems silly.
dwmw2rdieter: proposal: "join us in the 21st century"
rdieter:)
dwmw2if I came up with a proposal that we should avoid vowels, would you need a counter-proposal? Or just a "don't be silly" ?
warrenIt seems that we're rather split on this.  Perhaps easier to agree on disallowing for F9 and putting more discussion into it after?
spot+1 for ascii naming
dwmw2at the very least, we should reject this proposal on the basis that there is no good reasoning being explained
let them come back with a coherent reason for it
jwbi think we're at a net total of 2, 7 is needed to pass
bpeppleok, I see four '+1', and one '0', and three '-1'.
dwmw2disallowing for F9 makes sense. But _only_ for F9.
tibbsActually that seems to make little sense.
jwbdwmw2, to be fair, there might be good reasoning in archives or something somewhere
nottingwell, f9 and all prior releases
tibbsIt's obviously an issue for rel-eng as to wiether the infrastructure can handle things.
caillonjwb, good reasoning should be presented in the proposal, then.
dgilmoredwmw2: a development cycle with UTF-8 makes sense  so that if anything is broken  it can get fixed
jwbcaillon, that's why i asked for a summary
anyway, we digress
nottingi.e., no branches for epel, no branches for f8, f9 after f9 is released, etc.
tibbsThe other packaging guidelines do not generally present rationale.
Is FESCo planning to request rationale for all guidelines?
dwmw2notting: will, I was thinking more of infrastructure testing -- we don't really want to do it right now. If everything's working there's no fundamental reason we can't ship écolier-fonts for F9 and F8 at a later date, once we're sure
tibbs: no, only the stupid ones
tibbsBecause FESCo wishes to modify FPC procedure then it should say so.
nottingdwmw2: i suspect that the chances of everything working right now is pretty much zero
* spot notes that there is nothing preventing a package from Provides: écolier-fonts or Provides: 窓際族
warrentibbs, there might be legitimate technical reasons for this to be a pragmatic limitation today, but given how contentious it is (and the others are not) it might be reasonable to ask for rationale.
tibbswarren: That would be up to FESCo.
spotbut having the common-sense ability to deal with a package in a standard character set isn't "stupid"
dwmw2Proposal: Approve for f9 only; for later releases a proper rationale other than "we might have bugs and can't be bothered to check" should be presented and we'll reconsider.
warrendwmw2, +1
nirikspot: what about "- MUST: The spec file must be written in American English."
caillondwmw2, +1
tibbs"can't be bothered to check"?  It is not within FPC's ability to check.
That's a rel-eng issue.
jwbno
caillondwmw2, F9 and under, actually.
jwbthat's a fedora community issue
spotdwmw2: thats not the point of this draft
dwmw2tibbs: 'we' as in the Fedora project.
f13+1 on the propoasl.
caillonso nobody decides to try and branch for F8 ;-)
bpepple+1, I can live with dwmw2's compromise.
tibbs-1
spotit is not "we might have bugs and can't be bothered to check"
-1
jwb0
* spot is rather offended at dwmw2's tone and assumptions
f13I'm +1 to the ASCIINaming proposal from FPC that is.
warrenI am -1 to the ASCIINaming proposal from FPC that is.
tibbsspot: Get used to it, I guess.
warrenThis is very confusing
jwbagreed
warrenwe're all voting on slightly different things
=)
caillonwarren, that's how we roll
dwmw2let's do it again, for the original first and then for what I said?
jwbspot, can you give me (us?) a brief summary of the rationale then?
dwmw2or the other way round.
* nirik doesn't think either will pass, but ok.
spotThe purpose of ASCIINaming is so that people can reasonably perform package operations without opening a character map utility
jwberm... which people
dwmw2spot: I'd be happy with 'packages must provide an ASCII-only version of their name'
spotusers
caillonspot, why not require them to Provide: ecolier-fonts then?
dwmw2just as écolier-fonts Provides: ecolier-fonts.
spotcaillon: the default should be american english
this is the universal language used in the computing industry
f13dwmw2: because that doesn't have effect upon the file on the filesystem.
tibbsThat has all of the downsides with essentially none of the upsides.
spotjust like air traffic controllers have to use english with pilots
f13dwmw2: reverse it, and that's permittable.
warren窓際族 Provides uselessemployeesputnexttoawindowuntiltheyretire
tibbsAfter all, if you don't know what the transliteration is, how can you run --provides to find the transliteration?
dwmw2if you don't know what the package is in the first place, how can you... ? :)
tibbsyum search will tell you the package name.
f13dwmw2: ls, rpm -qpi foo
tibbsBut under your scheme it won't tell you the transliterated package name so that you can type it.
dwmw2cut and paste is a wonderful thing for those who don't know how to use their keyboards :)
jwbare there keyboards that don't have english characters on them?
spotdwmw2: are you seriously arguing that people should cut and paste the names of packages for operations?
nottingjwb: yes
warrendwmw2, to be fair, there are technical reasons why firefox GAVE UP on unicode domain names.  You really don't want to leave open the possibility of a package name being a unicode lookalike to some other package?
spotdo you really want people to actively hate using fedora?
jwbnotting, so wouldn't that require opening a character map for those people?
tibbsI don't have the ability to cut and paste in single user mode.  Do you?
caillonspot, they'll be using a GUI to install it anyway
dwmw2I was assuming that we wouldn't _actually_ have packages named like 窓際族
caillonPackageKit or whatever
spotcaillon: thats a rather short-sighted assumption
dwmw2It'd be stuff like écolier-fonts
tibbsI don't have packagekit in single user mode; do you?
dwmw2and people would type the 'obvious' transliteration and get away with it
f13dwmw2: that's teh problem
dwmw2just like people type naive and role and think it's correct
nottingjwb: oh, sorry, i read that as 'have non-english characters'. i don't know of keyboards that *don't* have any ascii
f13many times it' isn't obvious
tibbsPlease tell me the obvious transliteration of 窓際族
dwmw2I was assuming that we wouldn't _actually_ have packages named like 窓際族
caillontibbs, which is not the typical use case.  :)
spotdwmw2: kanji has no obvious transliteration. Neither does korean. Mandarin.
f13dwmw2: and unless we have a guideline against it, some asshat will bring a package named 窓際族 up for review
tibbscaillon: So you want to restrict it to typical use cases?
* spot will do it. i'll rename my packages.
tibbsCan you define those?
warrenspot, actually that isn't true, there are standard transliterations assuming you can read it to begin with.
jwbi see flaws in dwmw2's assumptions
dwmw2f13: don't we get to veto that on grounds of common sense?
jwbno
f13dwmw2: "common sense" is not a good guideline
jwbcommon sense is not a requirement for packaging software
f13dwmw2: because if it were really "common" it wouldn't have ever been brought up.
caillontibbs, no.  but an argument that most users will hate fedora for it doesn't hold water.
nottingjwb: because common sense is not a requirement for writing some of it?
jwbnotting, agreed.  we both know the package in question :)
warrenThere are likely VALID CURRENT TECHNICAL REASONS why this rule would be a good idea for F9, but I think it needs more study for the future, it would be premature to simply close the door entirely.
nottingjwb: oh, there are many
spotcaillon: every book ever written about Fedora (or RHL) defaults to the command line
bpeppleok, we seem to be at an impasse.  How do we want to resolve this?
spotthe command line has got to be usable by humans, without charmap tools or cutting and pasting
dwmw2bpepple: approve for F9, discuss later for the future
f13warren: every packaging guideline is up for review/change at any time
tibbsIf FESCo cannot approve the proposed guideline, then it simply doesn't become a guideline.
dwmw2test in rawhide after f9 by approving écolier-fonts as-is
f13warren: passing the guideline today doesn't mean we can't revisit and adjust at a later time.
tibbsAnd we'll go on with the reviewrs and CVS admins refusing to deal with those packages.
nottingtibbs: does that mean non-ascii are held until there is a guideline, or approved?
bpepplespot, tibbs: is dwmw2's proposal acceptable to you guy?
spotnotting: or transliterated
spotbpepple: no, it is not.
dgilmorespot: that makes sense
f13dwmw2: I'm not a big fan of breaking our distrubtion for /everybody/ just to see if there are bugs.
tibbsIf rel-eng folks want to test the infrastructure and report, that would be nice.
dwmw2it won't break for everybody. It won't even break for those who type 'yum install ecolier-fonts'.
spotthis is not a "hide from bugs" thing. it is a "ensure usability" thing.
jwbtibbs, you keep saying rel-eng
i don't think this is a rel-eng problem
tibbsOK, so who would do that, then?  Infrastructure?
warrenI suspect that even if we solved the URL downloading problems, this might end up being dangerous to support because of unicode lookalikes.
f13dwmw2: except we /already/ broke anaconda in multiple places by having utf-8 in Summary
jwbtibbs, QA?  the fedora community?
tibbsAll I know is that it's not anything I can even test.
dwmw2I'm happy to say "package names may be non-ASCII but must Provide an obvious ASCII transliteration which people can use instead"
f13dwmw2: and I'm happy saying the opposite.
spotdwmw2: didn't we already point out why that is backwards?
jeremyf13: that's mostl because newt is a turd pile
tibbsjwb: And just how in the hell does the fedora community get a UTF named file on the mirrors?
f13jeremy: yes, it's a pile of turd, but a clear example of things that will break.
tibbsI mean, someone actually has to test this stuff.  I don't have access.  Neither does almost everyone else.
warren30 minute mark
tibbsSomeone knows whether koji will handle 窓際族 as a package name.  I don't.
nottingthat's easy enough to test :P
* warren submits 窓際族 to koji...
jwbscratch build
spot窓際族 isn't usable by humans.
its as simple as that.
jwbSTOP
warrenspot, only about 1.5B humans...
f13and if we're really serious about it, and if we want to put resources behind testing this ttuff, we can put up a test koji instance
jwbjust vote on the FPC proposal again as is.
nottingi suspect mirrors will be a far far bigger problem than koji
spot+1
f13so that we can do test builds, and test composes.
dwmw2-1
bpepple+1
warren-1
tibbs+1 fpc proposal
f13+1 to fpc proposal.
nirik+1 fpc proposal
notting+1 on FPC. can be revisited later
jwb+1
that's 7
move on
* bpepple agrees with jwb.
jwboh, and what notting said about revisit
--- bpepple has changed the topic to: FPC Guidelines - http://fedoraproject.org/wiki/PackagingDrafts/Perl
spotguidelines can always be revisited/revised.
nirikfine with me, +1
tibbs+1
spot+1
notting+1 perl. well, perl guidelines. 0 or -1 on perl itself
bpepple+1 to perl proposal.
warren+1
dgilmore+1
bpeppleok, that's seven '+1'. the proposal has passed.
dwmw2er, this one also has multilib issues, doesn't it?
Requires: perl(:MODULE_COMPAT_%(eval "`%{__perl} -V:version`"; echo $version))
tibbsarch-specific perl modules will depend on the proper flavor of libperl.
dwmw2how?
tibbsThey link against it.
dwmw2nah, it doesn't work like that
oh, maybe it does
yeah, ok.
bpeppleok, if there's nothing else....
--- bpepple has changed the topic to: FPC Guideline - http://fedoraproject.org/wiki/PackagingDrafts/OpenOffice.orgExtensions
spot+1
notting+1
tibbs+1
dgilmore+1
bpepple+1
nirik+1
dwmw2+1
warren+1
bpeppleok, thats eight '+1', so that proposal has also been approved.
And I believe that's all we have from the FPC for this week.
jwbwhew
bpeppleok, let's move on...
--- bpepple has changed the topic to: FESCo-Meeting -- Final Release Schedule (Slip in conjunction w/ latest beta slip?) - all
bpeppleis wwoods about?
jeremyhe was not too long ago
f13This week we only had one day where rawhide wasn't installable IIRC, and that was due to a kernel change that wasn't fully tested.
jwbum, except today
f13jwb: ppc doesn't count (:
dwmw2and the day when the firmware packages with UTF-8 ate anaconda, although I suppose we could work around that
jwbf13, >:\
f13dwmw2: yeah, that's easily work aroundable, and only hits text installs
jeremydwmw2: that wasn't new this week.  and only hit text mode
f13hrm, we're supposed to be doing a snapshot today
jwbhow many in FESCo run rawhide?
f13but we know ppc will bust unless we pull in a newer x, which I think we can do
bpeppleI do.
* f13
dwmw2 does
jwb
tibbsI do.  But I just update it; I rarely try to install it.
jeremyf13: yeah... I've got a couple of livecds and have been testing some things (as a result of trying to fix bugs)
nottingjwb: run? yes. install? only on occasion.
* warren does
dwmw2I've done a bunch of installs this week on various machines
jwbnotting, i know that.
warrenHow many in FESCO yum update directly from static-repos?
* nirik does, except not on his main laptop yet
dgilmorewarren: i do
dwmw2not from the static repos. I yum update from my local rawhide mirror
tibbsI just use the mirrors, or pull from koji when I need something specific.
dwmw2my DSL line is too crap for anything else
bpepplewarren: I don't.  I just use the mirrors.
jwbi ask because if we had only few people running rawhide i'd sort of question our ability to vote on this
* wwoods is around *and* about
nirikI try and do installs for x86_64 base case every day that it's installable.
jwbbut it seems we have quite a few who do, so that's good
whitenoise warren wwoods
dwmw2don't know or care about i386/x86_64 but it looks good enough on ppc :)
bpepplewwoods: we are discussing whether we need to slip the release, and wanted to get you input for the qa side of things.
f13warren: I do
jwbwwoods, bring the pain!
wwoodshoorj
f13not slipping now doesn't prevent our ability to slip later if we really need it
but it would help us to plan things later accordingly.
wwoodsI haven't scrutinized the blocker list fully
f13the question is, do we really think we need to slip?
wwoodsthere's a lot of stuff on there that seems like it might already be fixed
f13(and in addition to, do we really thikn we have the ability to make that call ourselves)
wwoodsif it's *not*, and those are all still outstanding bugs.. then we'll probably need to slip
* bpepple leans toward not slipping.
wwoodsbut my current gut feeling from using rawhide myself and reading user test reports
is that we'll probably be OK for release by.. what, 22 April is the RC target?
f13yep
8'th is the final freeze
wwoodsI'd have to poll ajax / davej / jeremy / clumens / etc for their feelings on various major system components
but my highlevel gut feeling is: we could probably make it. probably.
f13Proposal: Get reports from Feature owners on thoughts of the 8'th final freeze, 22'nd RC and if they feel that's enough time for their feature.  Gather feedback, propose slip or not based on feedback.
(really what we should have done last week)
jwbdwmw2, i'm comfortable with where PPC is at.  you?
bpepplef13: +1.
dwmw2jwb: yeah
warrenDid spevack promise that we would be out by any upcoming event this time? =)
jwbwarren, we'd have to ask stickster
but i think, no
wwoodswarren: good question.. I'd really like to know if there're any hard deadlines
nirik+1 to f13's proposal./
jwb+1 to f13
warren+1
f13I know of no hard deadlines
I'd really like to be able to hand out F9 at Linux Fest Northwest which is the last weekend in April
wwoodsalso: I am going to be gone for the final freeze
bpeppleThough I believe marketing is pushing to have their ambassadors hold release parties, which would be effected by a slip.
sticksterThe summit is later this year so we're not really bumping against that.
f13but I haven't promised anything
jeremywary +1
jwbjeremy, why
wwoodsApr 7-11 I am completely unreachable because of family junk two years in planning
tibbs+1 to f13's proposal, but we need to decide this soon.
jeremyjwb: I have not been having a good week in terms of sitting down to try something and having it work, no problems
f13tibbs: yeah, could be done on list after feeback is gathered.
nottingbpepple: party on. release later!
f13poelcat: you around?
* poelcat lurking
notting knows of enough stuff he personally has to fix that he's wary
poelcatf13: behind on backscroll
f13poelcat: would you be willing to ping feature owners for feedback on their feelings of meeting the final freeze of the 8th and release candidate of the 22nd?
poelcat: we'd want feedback early next week to help decide a pre-emptive slip
jwbnotting, jeremy: are you filing blocker bugs?
poelcatf13: sure,  but wouldn't we just drop unfinished features?
nottingjwb: yes
spevackwarren: no, no hard deadlines.  I learned my lesson last year :)
poelcatf13: IOW are we saying some features *must* be done to ship?
f13poelcat: well, depends on if they need an extra day or so.  Sometimes it's hard to "drop" something that's tightly integrated but not quite done
jwbnotting, good
jeremyjwb: one better, I'm committing fixes
f13poelcat: no, we're asking for their opinions on being able to finish their features in a reasonable way by the freeze.
jeremyjwb: when I don't do that, I'm filing bugs
spot+1 to f13's proposal
* jwb bows before jeremy's ninja skillz
poelcatf13: okay, that helps for the wording of my email
individual feature owners or just to fedora-devel-announce?
warrenspevack, is that why we replaced you? =)
f13I'd go individual owners.
poelcatf13: will do
target EOB tomorrow
or sooner
f13spevack: you say that as if a dead horses head wound up in your bed next to you...
f13spevack: and I have nothing whatsoever to do with that if that did happen...
spevackf13: someone left a mountain of PPC DVDs in front of my apartment with a knife stuck in them
:)
f13spevack: now /that/ I may have had something to do with
* jwb frowns at spevack
EvilBobYeah if anyone needs any I have some FC6 PPC DVD here
f13lmacken probably had 50 or so in his cube when he started.  They were the welcoming committee
EvilBobthink i have about 50 left
spotjwb: come on, you know that won't kill them. ;)
f13it'll just anger them.
anywho
jwbspot, death would not garner me more users
f13bpepple: looks like we're all in favor of that proposal, and poelcat is off to get some feedback.
bpepplef13: agreed. we can move on.
poelcat: Is there anything in regard to Features we need to discuss today?
poelcatbpepple: no, i'm still detoxing from the past weeks ;-)
bpepplepoelcat: cool.
--- bpepple has changed the topic to: FESCo meeting -- Free discussion around Fedora
bpeppleok, anything folks want to discuss before we call it quits for today?
jwbso i suck and haven't done the updates proposals i promised
erm, bodhi proposals that is
i don't think i'll get to them before next week
bpepplejwb: np.  I think most of us are pretty busy right.
now.
jwb:)
bpepplealright, I think we can wrap the meeting up.
* bpepple will end the meeting in 60
bpepple will end the meeting in 30
bpepple will end the meeting in 15
bpepple-- MARK -- Meeting End

Generated by irclog2html.py 2.5 by Marius Gedminas - find it at mg.pov.lt!