Gentoo Logo
Gentoo Spaceship

Note: Due to technical difficulties, the Archives are currently not up to date. GMANE provides an alternative service for most mailing lists.
c.f. bug 424647
List Archive: gentoo-dev
Lists: gentoo-dev: < Prev By Thread Next > < Prev By Date Next >
To: gentoo-releng@g.o, gentoo-dev@g.o
From: Jeffrey Forman <jforman@g.o>
Subject: Meeting Notes - Release Engineering 4/14/05
Date: Thu, 14 Apr 2005 16:29:31 -0400
Attached is the notes of Release Engineering meeting.
Seemed pretty productive on our part. 

Devs: If you have not signed up for the devwiki yet, /topic
#gentoo-releng and ping me. I think the wiki is going to be a huge
headache saver for the next release. Sorry non-dev's, somethings must be
kept secret from the children ;)

For Release Engineering,

Jeffrey Forman
Gentoo Infrastructure
Gentoo Release Engineering
15:00 <@wolf31o2-work> let's do this... who's logging?
15:00 <@jforman> i can log
15:00 <@plasmaroo> I can.
15:00 <@plasmaroo> If Jeff /wants/ to do it, he can ;)
15:00 <@jforman> i got it
15:00 <@wolf31o2-work> heh... let's get started
15:01  * jforman sounds the horns
15:01  * kloeri waves
15:01 <@rocket> heh
15:01  * plasmaroo wonders whether wolf31o2-work just got lynched by work.
15:01 <@wolf31o2-work> nah
15:02  * jforman hears crickets
15:03 <@wolf31o2-work> I guess the first thing is pretty simple... what did we do right/wrong for 2005.0?
15:03  * plasmaroo goes back into guru meditation.
15:03 <@plasmaroo> Well, next time we need to check everything MD5 matches before release.
15:03 <@rocket> we made a cd is that right? :p
15:03 <@jforman> i think things went well, maybe a little communication glitches. i think the biggest thing was PR
15:03 <+Pylon> wolf31o2-work: Good: communication (during the second attempt).
15:03 <@wolf31o2-work> I think our communication was *really* poor at first...
15:03 <+gustavoz> positive: not rushing the release, time to fix stuff
15:03 <+Pylon> Bad: Delays and the store-preorder...
15:03 <@wolf31o2-work> yeah... we did quite well the second time around... heh
15:04 <@plasmaroo> Right, hopefully that'll be better now that the top-sekrit wiki infoTracker is up for 2005.1
15:04 <@jforman> i dont know how to figure out to set a sort "code freeze" where we stop adding stuff and only fix it. like things being late *cough* ppc *cough*
15:04 <+Pylon> jforman: ppc was earlier than ppc64.
15:04 <@jforman> err, ppc64, yeah. one of the fruity releases
15:05 <+tgall> yeah ppc64 was definitely late .. that's my fault entirely .. had one too many business trips at critical times
15:05 <+Pylon> And i wasn't used to release-building for ppc.
15:05 <@rocket> Documentation on the cds seemed to be an issue according to bugs.g.o
15:05 <@plasmaroo> Pylon: It still went extremely well :)
15:05 <@jforman> its cool. with the wiki i hope we can clear some of those things up.
15:05 <+tgall> tho for 2005.1 we're already working on a ppc64 release :-)
15:05 <+tgall> yes I like the wiki idea alot!
15:05 <@jforman> well we seem to be naming the things that are bad...lets name some things that went WELL
15:06 <+spy|work> can someone explain the wiki idea?
15:06 <@plasmaroo> Getting a kick ass release out.
15:06 <@plasmaroo> spy|work: SUre.
15:06 <@rocket> gensplash :p
15:06 <+tgall> catalyst was in very good shape .. least it felt that way to me
15:06 <@plasmaroo> spy|work: Basically, there's a status grid with everything important arch coordinators need to know.
15:06 <@plasmaroo> spy|work: They update things there themselves, and we know where everything is.
15:06 <@plasmaroo> spy|work: While whoever does the snapshot can post stickies with new snapshots and so forth.
15:07 <@rocket> I think wolf31o2-work worked very hard that was good for 2005.0
15:07 <@wolf31o2-work> ok... so we need to keep up communication (wiki)... not bow to outside pressure (store/users) and ensure we're confident in what we're releasing (nothing half-assed)... we also need to make sure that *everything* is done, ready, and tested before we stage *anything*
15:07 <+spy|work>
15:07 <+spy|work> kinda like that?
15:07 <+Pylon> jforman: Further good things: The ppc CD works on more machines than expected (and that untested).
15:07 <@jforman> spy|work: precisely
15:07 <+spy|work> except actually kept up to date
15:07 <@plasmaroo> jforman: Yep.
15:07 <@plasmaroo> spy|work: Yep.
15:08 <@jforman> wolf31o2-work: yeah, i think we need to workout a written down policy of how we deal with things regarding the store
15:08 <@plasmaroo> Yep.
15:08 <@wolf31o2-work> jforman: ok... are you volunteering?
15:08 <@wolf31o2-work> or should that be up to the trustees... since they're the ones with the agreement with the store, not us?
15:08 <@plasmaroo> Probably them.
15:08 <@jforman> wolf31o2-work: i'd rather stay out of the politics of it all. if i deal with the situation without needing a flame-retardant suit, by all means
15:08 <@rocket> I would say them
15:08 <@plasmaroo> We may have legal agreements.#
15:08 <+tgall> Pylon, next time sure wanna get chrp ppc32 hardware covered
15:08 <@jforman> okay, i can mention it to corey and see how to handle it
15:09  * jforman adds this to the wiki
15:09 <+Pylon> tgall: And OldWorld.
15:09 <@wolf31o2-work> ok... so we need to ask the trustees to come up with something regarding the store
15:09 <+tgall> Pylon, yeah good point
15:09 <@jforman> hey guys, i'm going to add something of a "minutes" to the wiki with action points
15:09 <@plasmaroo> Yep.
15:09 <@wolf31o2-work> we done with 2005.0? *grin*
15:09 <@wolf31o2-work> jforman: sweet
15:09 <@plasmaroo> wolf31o2-work: Yes.
15:09 <@plasmaroo> jforman: Action points?
15:09 <@wolf31o2-work> What can we do to help others in releasing that didn't release 2005.0?
15:10 <@plasmaroo> Find a better timeslot for the release, really.
15:10 <@jforman> plasmaroo: things we discussed that we'd lke to get done
15:10 <@plasmaroo> I think the MIPS and HPPA people were just ... busy.
15:10 <@plasmaroo> jforman: Ah, cool.
15:10 <@wolf31o2-work> I was thinking doing more like ppc did... have more than a single person responsible for the release per arch...
15:10 <@plasmaroo> HPPA is also really slow at building as GMSoft has slow hardware.
15:10 <+tgall> getting docs updated / onto the universal cd I'm a bit gray on yet how that process is supposed to work
15:10 <@jforman> wolf31o2-work: if we do that, who does coordination fall to?
15:10 <@plasmaroo> wolf31o2-work: It's harder to do that for rare architectures.
15:11 <@plasmaroo> wolf31o2-work: It'll work for say, PPC/x86 but not for say, HPPA.
15:11 <+agaffney> plasmaroo: he just got that C3600
15:11 <@plasmaroo> agaffney: Ah, cool.
15:11 <@wolf31o2-work> jforman: there's still a coordinator per arch... but that person shouldn't be solely responsible for the actual building... they just coordinate (how it is supposed to be anyway)
15:11 <@plasmaroo> agaffney: That'll help then.
15:11 <+tgall> wolf31o2-work, in the FYI catagory, the ppc and ppc64 projects will be merging over time ...  pylon and I haven't worked out how the next round of livecds for 2005.1 will get put together quite yet
15:11 <@jforman> wolf31o2-work: correct. makes sense with me
15:12 <@wolf31o2-work> tgall: yeah... I've heard that... I think it'll be a good move for all...
15:12  * jforman agrees
15:12  * plasmaroo nods
15:12 <@wolf31o2-work> next item...
15:12 <@wolf31o2-work> What other teams should we be coordinating with more closely?
15:12 <@wolf31o2-work> #1. docs
15:12 <@wolf31o2-work> heh...
15:12 <@wolf31o2-work> #2. PR
15:12 <@plasmaroo> Docs, definitely.
15:12 <@rocket> yeah
15:12 <+gustavoz> #3 desktop
15:12 <@plasmaroo> Does PR have enough staff?
15:12 <@jforman> PR i think was a big clusterfsck.
15:13 <@plasmaroo> Right.
15:13 <+Pylon> plasmaroo: Currently no.
15:13 <@jforman> someone on PR needs to have a list of "these are the sites we publicize to"
15:13 <@plasmaroo> jforman: SwifT posted one to -core
15:13 <@jforman> plasmaroo: nod, good point
15:13 <@wolf31o2-work> right... and with him having stepped down, we have no real outside PR person
15:13 <@plasmaroo> Right...
15:14 <@wolf31o2-work> jforman: think we could recruit jfmuggs for the job for the next release?
15:14 <+Pylon> wolf31o2-work: plate is our PR-man.
15:14 <@jforman> wolf31o2-work: if he's willing to do it. he can speak the good ameican
15:14 <@jforman> *amrican
15:14 <+Kugelfang> hey peeps... sorry, i fell asleep :-/
15:14 <@roger55> hi
15:14 <+spy|work> actually ulrich is pretty good
15:14 <+spy|work> oops Pylon said it
15:14 <@wolf31o2-work> Pylon: plate does an awesome job when it comes to the GWN... I think if he wants the job, he would be good for it...
15:14 <@plasmaroo> Ok, so we just need to speak to those people then.
15:15 <@plasmaroo> jforman: ^^ Action point
15:15 <@wolf31o2-work> I just haven't seen him step up and say that he'd do it
15:15 <+Pylon> wolf31o2-work: He is currently our single left PR-member.
15:15 <+Pylon> wolf31o2-work: And he has some ideas for that team.
15:15 <+Pylon> He couldn't do that alone.
15:15 <@wolf31o2-work> this actually is going intot he next question, which is: What should we do regarding PR (or the lack thereof)?
15:15 <+spy|work> i'm on the pr alias.
15:16 <@wolf31o2-work> should we assign someone to assist with PR at release time?
15:16 <@plasmaroo> Yep, that might work.
15:16 <@plasmaroo> So we'll need a PR contact.
15:16 <@plasmaroo> Who can then liase with the rest of PR or something.
15:16 <+Pylon> I offered my help to PR, but only for events.  That's enough to do.
15:16 <@wolf31o2-work> I think it would definitely help... John (zhen) used to do it when he was around...
15:16 <+tgall> certainly would help to get information into their hands
15:16 <+spy|work> i was kinda thinking swift was still on pr, just not the lead
15:16 <@jforman> i think we need to come up with a standard single release that we paste to all the slashdot/osnews/linuxtoday/etc sites. PR constructed
15:16 <@plasmaroo> slashdot is a mess.
15:16 <@plasmaroo> We might need to talk to their editors.
15:16 <@plasmaroo> Or try to.
15:16  * plasmaroo grins.
15:17 <+spy|work> i agree with that, jforman. the release needs to get looked at by more people.
15:17 <@wolf31o2-work> jforman: right... but we can't really control slashdot... they're pretty... eh... independent....
15:17 <@jforman> yeah
15:17 <@jforman> i threw up a  thought on the wiki.
15:17 <+Pylon> Try to focus on other news-sites.
15:17 <+blackace> are you going to mop it up?
15:17 <+Pylon> PR about 2004.3 was pretty well.
15:18 <@jforman> Pylon: we can add that into the "Release notes" on the wiki 
15:18 <+Pylon> But about 2005.0 I found quite nothing.
15:18 <@wolf31o2-work> ok... so we should try to come up with someone (whether it be an actual person on PR or one of us) to assure PR is handled for the next release and to help even out the load...
15:18 <@roger55> How about making the release news item a 'full story ' thing with a lot more information, linking to the various release notes?
15:18 <+tgall> I'd advocate someone have a foot in both releng and PR 
15:18 <+spy|work> whatever ends up happening, please filter the release through pr beforehand if nothing else
15:18 <@plasmaroo> spy|work: We did, and I think we got no feedback from them.
15:18 <@wolf31o2-work> roger55: that's a really good idea... I wrote the release stuff this time as an afterthought (since I'd actually forgotten it... heh)
15:18 <+spy|work> i don't recall seeing it
15:19 <+tgall> roger55, I like that idea ... allow all the arches to say what's new and sparklie in this release
15:19 <+Pylon> tgall: Don't look at me... ;-)
15:19 <+spy|work> ah, there's a two-paragraph thing from chris.
15:19 <@wolf31o2-work> spb: I sent it through PR... which is where plate grabbed it for the GWN...
15:19 <+seemant> spy|work: hey I have a question for you
15:19 <+spy|work> with responses from corey and sven, both on pr
15:19 <+seemant> spy|work: you mind if I bump xter-200 to stable?
15:19 <+spy|work> seemant: go nuts.
15:19 <+seemant> thanks sweetheart
15:19 <@plasmaroo> spy|work: Ah, ok, maybe I didn't see the responses as I'm not on pr@, heh.
15:20 <+Pylon> Okay.  As long as I don't have to do tactical manager and only Releng for ppc, and as I have a good contact to Ulrich, I can look into the PR liason.
15:20 <@jforman> wolf31o2-work: if you look on the wiki, in the "Release procedure" there is a point for getting PR all setup
15:20 <+spy|work> plasmaroo: you were CC'd
15:20 <@wolf31o2-work> ok...
15:20 <@plasmaroo> spy|work: Hrm, interesting...
15:20 <+spy|work> the release looked pretty good anyhow, nobody had an edit.
15:20 <@wolf31o2-work> the last 3 items I have really are all the same thing...
15:20 <+spy|work> confirmation that chris does speak the english language natively.
15:21 <@wolf31o2-work> heh
15:21 <@wolf31o2-work> what can we do to make 2005.1 better?
15:21 <@roger55> currently our release notes are a bit hidden behind the releng project pages. If we stuck some of the bugs with resolutions in a more verbose errata section that'll save users some trouble, should there be any issues at all with 2005.1 ;)
15:21 <@wolf31o2-work> ...or do we even really need to?
15:21 <+Pylon> wolf31o2-work: Better documentation.
15:21 <+Pylon> wolf31o2-work: Easier installation process.
15:21  * roger55 is lagging badly over the mobile phone link...
15:21 <+Kugelfang> wolf31o2-work: testing stages in between
15:21 <+Pylon> wolf31o2-work: Leave out stage2 and offer only two stages.
15:21 <+Kugelfang> wolf31o2-work: would be good to know that the profiles actually work
15:22 <@wolf31o2-work> roger55: ok... more prominent release notes/errata... sounds good...
15:22 <+spy|work> an obvious errata does sound useful
15:22 <@wolf31o2-work> Kugelfang: blameamd64... :P
15:22 <+Pylon> wolf31o2-work: Harder testing of the profiles in front of the release.
15:22 <+Kugelfang> wolf31o2-work: hehe :-)
15:22 <@wolf31o2-work> another thing... release process changes...
15:22 <@jforman> wolf31o2-work: i think we need to think about how subproducts want to get handled. (like hardened)
15:22 <+Kugelfang> Pylon: yeah, what's better than running stages 1 up to 3 once a month
15:23 <+Pylon> Kugelfang: Running also stage3.
15:23 <+Pylon> Kugelfang: Remember, with the 2004.3 release ppc had only issues with stage3 :-/
15:23 <+Kugelfang> ah, the release w/o ppc ?
15:23 <@plasmaroo> And stage3 is most used along with stage1
15:23 <@wolf31o2-work> jforman: definitely... the current situation is dismal...
15:23 <@plasmaroo> jforman: Yep.
15:23 <@rocket> I guess I would like to see us not have different cds for the store and for the mirrors ..
15:23 <+Pylon> wolf31o2-work: And another thing: Let's do the release within student's holidays.  It's stupid to do that within exam's week.
15:24 <+Pylon> Kugelfang: No, that was 2004.2.  But there were pam-issues due to an unproper built stage3.
15:24 <@jforman> Pylon: we can do that. take release dates into consideration with our younger population
15:24 <@rocket> I think that messed up tracking some the md5's .. and you never know if a bug in bugs.g.o might have been from different cd's ..
15:24 <@wolf31o2-work> Pylon: I don't have a clue what school schedules are like... we'll get to schedules later (in a few minutes)
15:24 <@roger55> wolf31o2-work, I'd like to see a wider testing pre-final material by users. we'd need to provide a howto for that though.
15:24 <+tgall> school? what's that?
15:24 <+Pylon> wolf31o2-work: Okay.
15:24 <@wolf31o2-work> rocket: yeah... well... the store isn't getting anything this time until we are *ready*
15:25 <@rocket> sounds good .. just wanted to raise the concern
15:25 <@wolf31o2-work> roger55: if you found a good subset of users... that would be awesome...
15:25  * Kugelfang points at the AMD64 Archtester
15:25 <+Pylon> roger55: Talking about store-CDs.  It would be nice to have some requirements for those CDs/DVDs.  I don't want to remember back to my discussion with drobbins and the ppc-CD.
15:25 <+Kugelfang> +s
15:25  * kloeri would be happy just to have a few users testing alpha.. :)
15:25 <@roger55> wolf31o2-work, if I have a decent internet connection at that time I'll recruit some form #gentoo and friends.
15:25 <@wolf31o2-work> koyou have me... :P
15:26 <@jforman> Pylon: thats going to be handled in our discussion with the trustees about our relationship with the store
15:26 <@wolf31o2-work> roger55: cool... =]
15:26 <+Pylon> jforman: Good and thanks.
15:26 <@rocket> possibly better use of /experimental .. to get more testers?
15:26 <+Pylon> kloeri: Build a release for a DEC3000 and I can test :-P
15:26 <@wolf31o2-work> now... here's something I'd really like to disccus... release guidelines... the ones we have are a bit stale... I know i haven't touched them since I started working on releng...
15:26 <@jforman> Pylon: are you on the wiki? if not, devwiki.g.o, sign up and i can bump your privs
15:26 <+kloeri> Pylon: heh, I'll see what I can do :)
15:26 <+Pylon> rocket: Yeah.  ppc will make use of experimental this time.
15:26 <@jforman> wolf31o2-work: guidelines as in? (got a link)
15:27 <@wolf31o2-work> everything pre-release *must* use /experimental... I think I'm going to have -infra enforce that pretty heavily...
15:27 <+Pylon> jforman: I'm already on, but didn't looked into it during the last three weeks.
15:27 -!- ratmice [bosshog@...] has left #gentoo-releng []
15:27 <+spy|work> just make release read-only when it's not around release time
15:27 <@jforman> Pylon: ahh okay, releng content has exploded ;)
15:27 <@roger55> rocket, the issue with testers is they need a guide on what to test for and how to report bugs and where. along with info on what to look out for in changes of the install procedure/livecd options...
15:27 <@wolf31o2-work>
15:27 <+Kugelfang> jforman: ahm, can you bump my privs for Releng please ?
15:27 <+Pylon> wolf31o2-work: Don't laugh.  ppc is about to release installation diskettes.
15:27 <@wolf31o2-work> Pylon: nice...
15:28 <@rocket> well that should probably go into the /experimental folder along with the files they download .. 
15:28 <+spy|work> i didn't realize catalyst could make floppies yet. =P
15:28 <+Pylon> That's part of our OldWorld strategy.  Some of the older PowerPCs can't boot from CD directly...
15:28 <@jforman> Kugelfang: done
15:28 <@rocket> spy|work: where did you find that?
15:28 <+Kugelfang> jforman: thx :-)
15:28 <+Pylon> Oh, yes.  Floppies.  I nearly forgot that term ;-)
15:28 <@wolf31o2-work> guess... back on-topic, please
15:28 <+Pylon> spy|work: They will be hand-made.
15:28 <+gustavoz> Pylon: can't they do netboot instead?
15:28 <@wolf31o2-work> guys
15:28 <+Pylon> wolf31o2-work: What's the topic?
15:29 <+spy|work> vapier had some floppy code for it, but incomplete
15:29 <@wolf31o2-work> now... here's something I'd really like to disccus... release guidelines... the ones we have are a bit stale... I know i haven't touched them since I started working on releng...
15:29 <+Pylon> gustavoz: Alternative installation instructions.
15:29 <@jforman> hey guys, let's try and stay on topic with chris
15:29 <+gustavoz> Pylon: sure, but given that you can't get decent blank floppy media it's worth looking
15:29 <@jforman> gustavoz: Pylon: take it to #-ppc please
15:29 <@plasmaroo> Ok, well the main bottlenecks with the current process are PR, release and partially building.
15:29 <@plasmaroo> Building should be improved for 2005.1 hopefully; we will see as we go.
15:30 <@plasmaroo> PR and the release is what needs focusing on, IMO.
15:30 <+Pylon> wolf31o2-work: Well.  They were a vague help on this release.  But there are some issues with them, agreed.
15:30 <@plasmaroo> s/Building/Coordination/
15:30 <@jforman> wolf31o2-work: i think we need to come up with some direct "if you are testing, this is what you need to look for" type things
15:30  * jforman adds to wiki
15:30 <@wolf31o2-work> Pylon: ok... what could be improved? I plan on really revamping them...
15:30 <@roger55> jforman, where's that wiki (I came late)?
15:30 <+kloeri> roger55: devwiki.g.o
15:31 <@roger55> thanks.
15:31 <@jforman> roger55: register can i can bump your access
15:31 <+kloeri> roger55: sign up and have jforman bump your access
15:31 <@rocket> first I think a link to them should be put in the topic for this channel 
15:31 <@wolf31o2-work> on-topic people
15:31 <+spy|work> guess the guidelines are pretty good =)
15:31 <@jforman> wolf31o2-work: i'm adding these thoughst to the wiki, wrt guidelines
15:31 <+Pylon> wolf31o2-work: Hmm.  I think they need a better coordination map.  Listing responsible groups/persons, like for the documentation, or just where you have to upload your built data etc.
15:31  * agaffney watches wolf31o2-work try to herd all the small children running around the channel
15:32 <@wolf31o2-work> Pylon: ok... that would help....
15:32 <@plasmaroo> Pylon: Right, hopefully we'll be moving all that info into the wiki so it should be easier, and that would be quite useful to have too.
15:32 <@wolf31o2-work> I was also thinking some hard requirements for the release.... like... must have X built by Y... must communicate Q with developer R.... that sort of thing...
15:32 <@plasmaroo> Hard requirements cause trouble.
15:32 <+Pylon> And list things that are possible and what not (like, that you can't unpack stuff on toucan, and re-build a CD with mkisofs...)
15:33 <@jforman> wolf31o2-work: i agree, which plasmaroo's table on the wiki should take care of. at least partially
15:33 <@wolf31o2-work> no... they keep me from freaking out and killing all of you
15:33 <@plasmaroo> We may needed harder requirements, but not rock-solid ones.
15:33 <@plasmaroo> wolf31o2-work: Killing people is my job, remember?
15:33 <+Pylon> And probably we should add to the guidelines a kind of common profile.  Like, what should be supported and what not.
15:33 <@plasmaroo> Right.
15:33 <+spy|work> yeah, prolly add a guideline to the list about getting in touch w/ PR
15:33 <@plasmaroo> And when to deprecate, etc.
15:33 <+Pylon> So that we have a little bit convergence bitween the different arches.
15:33 <+spy|work> and any other projects releng should be communicating with
15:33 <+Kugelfang> Pylon: thanks for reminding me Lars :-)
15:33 <@wolf31o2-work> oooh... deprecation... that's a good one I didn't think of...
15:34 <+Pylon> Currently some smaller arches do their own thing.  Still support devfs, though the major arches are on udev already.
15:34 <@rocket> I think the arches should be as close to identical as possible ...
15:34 -!- so|home [~so@...] has joined #gentoo-releng
15:34 <@wolf31o2-work> they are... only sparc is not and that's for technical reasons...
15:34 -!- mode/#gentoo-releng [+v so|home] by wolf31o2-work
15:34 <+Kugelfang> i have an idea i want to discuss with the amd64 people at the next meeting regarding the profiles
15:35 <+Kugelfang> i'd like to have a profiles ChangeLog, a mandatory one
15:35 <@rocket> well the only concern I had is making sure people are using official type specs ...
15:35 <+Pylon> It was a good help for me that I could compare the ppc-scripts to the x86-scripts.  Otherwise I would have forgotten some applications or some settings.
15:35 <@wolf31o2-work> something like upgrades should be clean and not require a phd to perform?
15:35 <+spy|work> some of the startups seemed to be a bit different irt the motd, the boot options screen, etc.
15:35 <@rocket> hopefully that can be taken care of by catalyst more in the future .. but still ..
15:36 <@wolf31o2-work> Kugelfang: oh yes... all you people need to add bugs you resolve regarding the release to the ChangeLog... if you do not have access to write to gentoo/xml/htdocs/proj/en/releng, then harass pylon
15:36 <+Kugelfang> wolf31o2-work: hehe, this one needed the script
15:36 <+spy|work> wolf31o2-work: here's a good one: you can't release until your specs are in cvs. =P
15:36 <@roger55> wolf31o2-work, can we include specfiles on the cds by default?
15:36 <+Kugelfang> wolf31o2-work: uh, we have a releng changelog ?
15:36 <@wolf31o2-work> Kugelfang: yes...
15:36 <+Kugelfang> great, this is the first time i get to know about it :-/
15:36 <+Pylon> wolf31o2-work: Every other devrel-member is allowed to grant access to that releng-path at CVS.
15:37 <+Pylon> I don't want to be that single-point-of-failure ;)
15:37 <@plasmaroo> Pylon: s/Devrel/recruiter/
15:37 <+spy|work> Pylon: devrel can change cvs permissions? that's the first i've heard.
15:37 <@plasmaroo> No, only recruiters.
15:37 <@plasmaroo> ... unless something is new.
15:37 <@jforman> hey guys, stay on topic
15:37 <+Kugelfang> Pylon: would you please ?
15:37 <+Kugelfang> :-)
15:37 <@wolf31o2-work> spy|work: that's really a good one... one of those rules... liek I said... I think certain things need to be added to the guidelines... "requirements"... things like... packagecd... specs... kernel configs...
15:37 <+Pylon> Ehm, yes, recruiters.  Just ask in #gentoo-infra
15:37 <+Kugelfang> kk
15:37 <@wolf31o2-work> roger55: I would prefer not have the specs on the CD, there's a package for them already... dev-util/livecd-specs
15:38 <+kloeri> hard requirements would be a big help for me at least - I tend to be so busy that I forget half the stuff I'm supposed to do :(
15:38 <+spy|work> yeah, maybe you could get the guidelines into a "release checklist" format even more
15:38 <@wolf31o2-work> kloeri: that's why we're making up a checklist... so you just follow them from start to end and you've fulfilled all the guidelines
15:38 <@wolf31o2-work> spyderous++
15:38 <@wolf31o2-work> exactly...
15:39 <@roger55> wolf31o2-work, if catalyst would add them one could be sure that those are the specs the cd was built with. 
15:39 <+spy|work> if you put 'em on the wiki, people could even check them off.
15:39 <+spy|work> i have a feeling we already talked about this.
15:39 <@plasmaroo> Right.
15:39 <@wolf31o2-work> yeah...
15:40 <@wolf31o2-work> roger55: and fsscripts? or any other files that might be used? I think it's really too much to ask... much easier to throw it in portage...
15:40 <+spy|work> now we just need the xsl translator guidexml<->wiki format
15:40 <@jforman> wolf31o2-work: dont forget spy|work's email about the direction of catalyst
15:40 <@jforman> spy|work: right now the devwiki is only for dev's
15:40 <@wolf31o2-work> jforman: that's after this last one...
15:40 <@wolf31o2-work> release dates...
15:41  * Pylon proposes late July or August.
15:41 <@wolf31o2-work> the tentative schedule for 2005.1 is the end of July... should we try to make that, or push it back?
15:41 <+tgall> for 2005.1?
15:41  * plasmaroo agrees.
15:41 <@wolf31o2-work> tgall: correct
15:41 <@jforman> push it back
15:41 <@roger55> september.
15:41 <+Pylon> September is again a problem.
15:41 <+tgall> I'm ok with end of July .. only slight concern it's close to OLS  that's July 20-23
15:41 <@rocket> push i think .. 
15:41 <@jforman> is there anything we are rushing to release? or can we push it back a bit and get in some "cool features" ?
15:41 <+Pylon> There are ofter exams for the students.
15:41 <+spy|work> seems like july/august and january are reasonable times with class schedules.
15:41 <+spy|work> if you could time it for the two lwe's, that'd be cool
15:42 <@jforman> spy|work: lwe sf is like 1st week of august iirc
15:42 <@wolf31o2-work> spy|work: that was my original intention...
15:42 <@jforman> wolf31o2-work: which means you stop development in early july :/ which is 3 months from now.
15:42 <@wolf31o2-work> jforman: there won't be anything "must have" for 2005.1, which is why I don't see a point in postponing...
15:42 <@jforman> wolf31o2-work: true, good point
15:42 <+Pylon> Well..  We in Germany are already used to provide old install-CDs at LinuxTag -- that's the reason why we build our specialised own CDs ;)
15:43 <+spy|work> it's just a snapshot, so the only must-haves are on the livecd side.
15:43 <@wolf31o2-work> 2005.1 should be a 2005.0 refresh... newer software... less bugs... ;]
15:43 <@jforman> more chicks
15:43 <+spy|work> figure out how long it's gonna take to get your X cd's working, release a month after
15:43 <+tgall> for ppc64, hoping that 2005.1 feature full working multilib
15:43 <@wolf31o2-work> spy|work: they work now... ask jforman... ;]
15:43 <+spy|work> wolf31o2-work: so catalyst cvs can build them, and it's actually documented?
15:44 <+Pylon> spy|work: Look at those fizzlewizzle-CDs.  They are a good starting point for X-LiveCDs.
15:44 <@wolf31o2-work> documented? hahaahahahaha
15:44 <+spy|work> it's not working until it's documented
15:44 <@wolf31o2-work> oh yeah...
15:44 <+agaffney> spy|work: catalyst documentation is an oxymoron
15:44 <@wolf31o2-work> jforman: add a note for me/rocket to update the documentation
15:44 <@jforman> done
15:44 <+spy|work> agaffney: so perhaps that should be a releng goal.
15:44 <@rocket> bah no docs .. :p
15:44 <+spy|work> get zhen's howto finished, and add the new crap to the reference
15:44 <@wolf31o2-work> the problem is that some of catalyst just doesn't make sense and is hard to follow without a good set of specs to look at...
15:45 <@wolf31o2-work> ok... so what did we decide on release date...
15:45 <+spy|work> wolf31o2-work: wouldn't that suggest it needs more documentation rather than zero?
15:45 <+tgall> yes very true .. example specs go a LONG way and keep people on the right path
15:45 <@wolf31o2-work> spy|work: I'm not a documentation guy... zhen was.... heh
15:46 <+spy|work> wolf31o2-work: then find yourself a documentation guy, and tell him how stuff works. =)
15:46 <@plasmaroo> I'm a doc guy but don't know we need docs ;)
15:46 <@wolf31o2-work> ok... how about this... yay or nay... 2005.1 end of July...?
15:46 <@wolf31o2-work> yay
15:46 <@plasmaroo> Yay
15:46 <+spy|work> yea
15:46 <@rocket> yay
15:46 <+kloeri> yay
15:46 <@roger55> y
15:46 <@jforman> sure why the hell not, i'm not the one building it ;)
15:46 <+Pylon> yep
15:46 <@plasmaroo> Yay it is :)
15:46 <@wolf31o2-work> ok... sounds good...
15:46 <@jforman> wolf31o2-work: does that include livecd ?
15:46 <@jforman> err Xlivecd
15:46 <+tgall> 2005.1 : yea
15:47 <@wolf31o2-work> jforman: experimental, yes... it won't be official for a while...
15:47 <@wolf31o2-work> ok... last question(s) from spyderous
15:47 <+agaffney> wolf31o2-work: I'll say yey but don't expect a 1.0 of the installer
15:47 <@rocket> just dont tell drobbins its end of july  :p
15:47 <@wolf31o2-work> catalyst/genkernel... where are they going...
15:48 <@plasmaroo> spy|work: Regarding genkernel, there's a lot of changes that have gone in for 3.2.0 which will be out Real Soon Now; e.g. initramfs support, etc.
15:48 <+spy|work> plasmaroo: one thing i'd like to see is ability to replace linuxrc w/ a custom one.
15:48 <@rocket> hrmm well should we use catalyst cvs for 2005.1 or use catalyst 1?
15:48 <+spy|work>  /etc/genkernel/linuxrc or something
15:48 <@plasmaroo> spy|work: That's going to be doable with an initramfs override.
15:48 <@plasmaroo> spy|work: Instead of giving options to replace $file, you just inject an override cpio with file.
15:48 <@wolf31o2-work> rocket has been doing some amazing things for catalyst 2.0 (cvs HEAD)... I am planning on simplifying the spec files for official releases by building more of the data into catalyst itself...
15:48 <@wolf31o2-work> rocket: it'll be catalyst 2, most likely....
15:49 <+Pylon> At least they worked for ppc.  But I guess, I have to add specialised kernels manually instead building them with genkernel.
15:49 <@plasmaroo> rocket: So that override patch will be useful.
15:49 <+tgall> rocket, I'd say use a catalyst that works :-)
15:49 <@rocket> the override will grab a directory actually and build the cpio on the fly
15:49 <@jforman> the big goal...a catalyst even jforman can use to build Xlivecds...that is the ultimate goal
15:49 <@plasmaroo> tgall: 2.0.0 has been great so far in my testing.
15:49 <+tgall> plasmaroo, tried that on a ppc64 box yet ?  ;-)
15:49 <@wolf31o2-work> one thing that will be much more important with catalyst 2.0 is livecd/type... the gentoo-release-* types will do quite a bit of the work for you... ;]
15:49 <@plasmaroo> tgall: No ;)
15:50 <+Kugelfang> wolf31o2-work: ahm, yay for end of july :-)
15:50 <@rocket> wolf31o2-work: should I go over a list of changes for catalyst 2 now?
15:50 <@wolf31o2-work> also, since we brought rocket on, catalyst and genkernel are tied together much closer... so it allows us to do quite a bit more that we simply couldn't do before...
15:50 <@rocket> and genkernel
15:50 <@wolf31o2-work> rocket: go for it...
15:50 <+tgall> plasmaroo, probably something to get nailed down in the next month then ... 
15:50 <+Pylon> I want pre-fetching of sources, so that catalyst can even work without network.
15:50 <@plasmaroo> tgall: Yep.
15:51 <@rocket> ok much improved is the kernel cache for genkernel
15:51 <@wolf31o2-work> Pylon: file a bug... :P
15:51 <@rocket> Pylon: post a bug request
15:51 <@rocket> kernel cache should work even if you change the gensplash them so you can leave it on
15:51 <+Pylon> Okay.
15:52 <@rocket> kernel cache will detect USE,EXTRAVERSION,and kernel .config changes and invalidate the cache if any of those differ between runs
15:52 <+agaffney> rocket: does kerncache handle the post-kernel modules better now (does it get their initscripts too)?
15:52 <+Pylon> Nice.
15:52 <@rocket> all post-kernel stuff are handled as part of the kname
15:52 <+Kugelfang> FYI: i just posted an enhancement bug to catalyst for automatical check and download of a (release-)snapshot by specifying an URL instead or additionally to a snapshot-date
15:53 <+Kugelfang> BUG #89124
15:53 <+GenBot> Kugelfang: Bug nor, P2, All, rocket@g.o, NEW, [REQ] Snapshot download on demand
15:53 <@rocket> any agaffney so it should be handled correctly
15:53 <+agaffney> rocket: that's all I wanted to hear :)
15:53 <@rocket> there is a separate pkgcache for each kname .. and those get installed every time genkernel runs against that kname
15:54 <+spy|work> plasmaroo: oh, i'd like a way to disable or clear the package cache for genkernel. could it just stick stuff in portage's packages instead? i had a hell of a time tracking it down after it built some broken ones.
15:54 <@rocket> you can find the kernel cache inside packages/arch/gk_binaries/kanme
15:55 -!- cshields [~cshields@...] has joined #gentoo-releng
15:55 <@rocket> spy|work: turn of kerncache in /etc/catalyst.conf and its disabled
15:55 -!- mode/#gentoo-releng [+v cshields] by wolf31o2-work
15:55 <+cshields> (thx) meeting still going?  I forgot about it  :|
15:55 <+agaffney> cshields: yep
15:55 <+spy|work> rocket: so kerncache also stops cache of busybox/devfsd/etc?
15:55 <@rocket> spy|work: actually we can discuss that .. that might not be possible yet
15:55 <+spy|work> that's obscure
15:55 <@wolf31o2-work> cshields: we're finishing up... but the open floor is still coming up... unless you've got something specific you wanted to ask?
15:56 <@rocket> spy|work: I *think* so but I have to test that
15:56 <@jforman> cshields: we (chris and i) can talk to you later about how the trustes "handle" the relationship with the store
15:56 <@jforman> wolf31o2-work: ^^^^
15:56 <@rocket> ok another big catalyst2 change
15:56 <+cshields> wolf31o2-work: open floor is what I'm interested in
15:56 <@rocket> archscript/runscript is gone
15:56 <+cshields> jforman: I'll comment on that later thx  :)
15:56 <@wolf31o2-work> spy|work: the original (catalyst 1.1.x) kerncache was an afterthought thrown in by zhen for his own personal use... it was never meant to really be used in its current form...
15:57 <@rocket> kerncache may need some tweaking but hopefully its better
15:57 <@wolf31o2-work> ok... so anything else or can we open the floor?
15:57 <@rocket> wolf31o2-work: want to talk about stage4 support?
15:57 <@wolf31o2-work> rocket: nah... we'll surprise them with that one when you write the docs for it... *cough*
15:57 <@rocket> catalyst 2 has much better support for autoresume
15:58  * Kugelfang gone now... have a paper to finish :-/
15:58 <@wolf31o2-work> thanks Kugelfang 
15:58 <@rocket> portage_overlay support has been added for testing new ebuilds in livecd-stage1 and livecd-stage2
15:58 <+spy|work> i think you guys shouldn't be allowed to release new versions until their functionality is documented.
15:59 <@wolf31o2-work> spy|work: we'd never release new versions, then... heh
15:59 <@rocket> catalyst2 prints the path to the snapshot initial stage along with its md5sum so you can make sure your using the correct image
15:59 <+spy|work> wolf31o2-work: some incentive to find a documenter or do it yourself, then
15:59 <+spy|work> all the code in the world is useless if nobody but you knows what to do with it
16:00 <@wolf31o2-work> spy|work: and I'm actually working on updating all of the example specs right now... so they'll be good documentation for most, if not all, of the options
16:00 <@rocket> and then alot of internal cleanups/simplification
16:00 <@rocket> I *think* that should cover most of it
16:01 -!- z1nOnly [~nick@...] has joined #gentoo-releng
16:01 <@rocket> and a super secret stage4 target that wolf31o2-work doesnt want me to talk about :p
16:01 <@wolf31o2-work> *grin*
16:01 <+Pylon> stage4 == emerge -e world
16:01 <+Pylon> It's so easy ;)
16:01 <@jforman> alright guys...we're opening it up to the world. that means non-devs can ask questions of us in here, discuss ideas they have, etc etc
16:01 -!- mode/#gentoo-releng [-m] by ChanServ
signature.asc (This is a digitally signed message part)
Lists: gentoo-dev: < Prev By Thread Next > < Prev By Date Next >
Previous by thread:
New developer: Robert Paskowitz (r2d2)
Next by thread:
Meeting Notes - Release Engineering 4/14/05
Previous by date:
Re: reply-to munging
Next by date:
Meeting Notes - Release Engineering 4/14/05

Updated Jun 17, 2009

Summary: Archive of the gentoo-dev mailing list.

Donate to support our development efforts.

Copyright 2001-2013 Gentoo Foundation, Inc. Questions, Comments? Contact us.