Summary from yesterdays EPEL SIG meeting

Thorsten Leemhuis fedora at leemhuis.info
Thu Apr 5 15:39:36 UTC 2007


= Meeting 20070405 =

[[TableOfContents]]

== Attending ==

>From the Steering Committee:

 * mmcgrath
 * nirik
 * quaid
 * thimm
 * thl

Active on the rabble seats:

 * che
 * entr0py

== Summary ==

 * RHEL5 final on the builders -- did still not happen due to legal
problems, that hopefully get solved soon

 * Mass rebuild of EPEL5  -- thl prepared a page on the wiki to try the
voting-via-the-wiki with this topic; details got send to the list in between

 * the mass rebuild discussion drifted of into the repotag issue. Some
discussions how to actually realize a repotag if we want one. Abusing
dist would result in a repo where only some packages use the disttag, as
using dist is optional in Fedora.

 * entr0py> | any comments on my request for brp-python-bytecompile?  ->
thimm> | entr0py: brp-python-bytecompile is currently under discussion
for other issues, too ;  So an improved version may appear in
*-rpm-config  which would fix your issue as well

 * RHEL5's $RELEASEVER translates to 5Server instead of 5 ->  the plan
is to hardcode 5 in the repo file to avoid problems

== Full Log ==

{{{
00:00 <       thimm> | epel meeting?
00:00 <     entr0py> | here
00:01 <         thl> | hi thimm entr0py
00:01              * | nirik was just going to ask.
00:01            --- | thl has changed the topic to: EPEL meeting
00:01              * | thl looks closer at his clock
00:01 <         thl> | I was in wokring in the wiki and lost time
00:01 <         thl> | ping mmcgrath
00:02 <         thl> | ping quaid
00:02            --> | sharkcz (Dan Horak)  has joined #fedora-meeting
00:02 <         thl> | dgilmore is busy, he can't join us today
00:02 <         thl> | mmcgrath is in phx iirc, so probably busy with
other stuff as well
00:02 <       quaid> | oh yeah!
00:03 <         thl> | hi quaid
00:03 <       quaid> | hi thl
00:03            --> | Kylixen (Kylixen)  has joined #fedora-meeting
00:03 <         thl> | so, some infors first
00:03 <         thl> | RHEL5 still not on the builders
00:03 <         thl> | there were some legal problems that stopped the
work for some days
00:04 <         thl> | but the problems gfot sorted out
00:04            --> | che (Rudolf Kastl)  has joined #fedora-meeting
00:04 <       nirik> | centos5 isn't out yet is it?
00:04 <    mmcgrath> | pong
00:04 <    mmcgrath> | I actually haven't left yet.
00:04 <       thimm> | what legal problem?
00:04 <         thl> | nirik, end of this week / beginning of next week
afaik
00:04              * | skvidal nods at thl
00:04 <         thl> | thimm, well, I think it was no real problem
00:05 <         thl> | thimm, someone feared a problems and asked legal
afaik
00:05 <         thl> | and everything got sorted out
00:05 <         thl> | I don't know any details about it
00:05 <    mmcgrath> | I sent a follow up email this morning regarding
RHEL this morning.
00:05            <-- | amitphukan has quit ("Leaving (আজিলৈ আহিলো)")
00:05 <    mmcgrath> | Haven't heard back.
00:05 <       thimm> | Is there anyway I can help with getting RHEL5 on
builders?
00:05 <         thl> | mmcgrath, afaics someone just needs to find the
time to install rhel5
00:06 <         thl> | mmcgrath, is that correct?
00:06 <    mmcgrath> | OH, are you talking about running RHEL5 on the
builders or having a RHEL5 repo for the mock to pull from?
00:06 <         thl> | mmcgrath, sorry,
00:07 <       nirik> | epel building will also be moving to koji this
weekend with fedora I imagine?
00:07 <         thl> | mmcgrath, I just meant: isntall a RHEL5 repo for
mock
00:07 <         thl> | nirik, I suppose so
00:07 <       thimm> | Hm, I have a RHEL5 repo for mock ...
00:07              * | skvidal does too
00:07 <       thimm> | I could rsync it somewhere
00:08 <     skvidal> | thimm: that's where legal gets cranky
00:08 <    mmcgrath> | thl: We actually have some RPM's there, the issue
is whether or not we have RH's thumbs up to use them.
00:08 <       thimm> | skvidal: but we have licenses for epel use, or?
00:08 <         thl> | mmcgrath, I thought we had the "thumbs up" now?
00:08 <    mmcgrath> | nope
00:08 <         thl> | :-/
00:09 <    mmcgrath> | Thats the email I sent this morning.
00:09 <       thimm> | mmcgrath: what is the problem?
00:09 <    mmcgrath> | Yep.
00:09 <    mmcgrath> | At first people thought we were going to be
distributing RHEL5 through Fedora (which caused WAY more people to get
involved then needed)
00:09 <    mmcgrath> | When I finally educated legal about what mock
does they were fine with it but worried that non RH employees would have
access to the RPMs.
00:10 <    mmcgrath> | I quelled that fear and now I'm literally waiting
for the person with the initial concern to have one final discussion
with Legal (AFAIK anyway)
00:10 <     skvidal> | ooo
00:10 <     skvidal> | yes
00:10 <    mmcgrath> | thats where its at now.
00:10 <     skvidal> | let's distribute rhel5 through fedora
00:10 <     skvidal> | that'd be fun
00:10 <         thl> | mmcgrath, is there any way we can speed solving
this porblem up?
00:10 <         thl> | mmcgrath, can quaid help?
00:10 <    mmcgrath> | thl: use CentOS.
00:11 <    mmcgrath> | This is why I sent that email earlier.
00:11 <    mmcgrath> | Max is involved so I don't think quaid can help.
00:11              * | quaid nods
00:11 <    mmcgrath> | This is one of those things where I've been
sending 2 emails a week and people just don't think its much of a priority.
00:11 <       thimm> | mmcgrath: Is this about getting licenses?
00:11 <    mmcgrath> | Though, and I really mean this, I'm pretty
hopeful that it will happen.
00:12 <    mmcgrath> | thimm: what licenses?
00:12 <       thimm> | For RHEL5
00:12 <    mmcgrath> | We're waiting on a phone call between two people,
one of which is in Phoenix and I'll be meeting with tonight and tomorow.
00:13 <    mmcgrath> | thimm: I'm not sure.  Because technically I don't
think we need a license to posess the RHEL rpms.
00:13 <         thl> | mmcgrath, okay, so I#d say let's hope and pray
for now
00:13 <    mmcgrath> | And even still we have 150 licenses to use for
Fedora (though there's strict rules about 3rd party stuff)
00:13 <         thl> | hopefully that stuff will sort out somehow over
the next few days
00:13 <    mmcgrath> | either way, the actual RHEL RPM's will never be
public.
00:13 <    mmcgrath> | thl: I'm hopeful... seriously.
00:13 <    mmcgrath> | :)
00:14              * | thl will move on if that's okay for everybody
00:14            --- | thl has changed the topic to: EPEL Meeting --
Mass rebuild of EPEL5
00:14 <         thl> | I was preparing a page in the wiki
00:14 <         thl> | to actually try voting via the wiki
00:14 <         thl> | and see how it works
00:15 <         thl> | not ready yet
http://fedoraproject.org/wiki/EPEL/SteeringCommittee/Voting
00:15              * | mmcgrath is never afraid to try new things.
00:15            --> | ClausReheis  has joined #fedora-meeting
00:15            --> | ClausReheis is Claus Reheis
00:15 <         thl> | shall we discuss this thing here again
00:15 <         thl> | we discussed it already once
00:15 <         thl> | and I know, I'm the one that changes his opinion
after the last meeting
00:15 <       nirik> | the voting? or the mass rebuilding?
00:15            --> | XulChris (Christopher Stone)  has joined
#fedora-meeting
00:16 <         thl> | nirik, that we want to do a mass rebuild is
settled afaics
00:16 <       thimm> | I still don't like forking specfiles
00:16 <         thl> | the question is how:
00:16 <         thl> | - delete and rebuild
00:16 <         thl> | - add a .1 top the spec files and rebuild
00:16 <       thimm> | - add a repotag and rebuild?
00:17 <         thl> | repotag is a different point
00:17 <       thimm> | No specfile forks with repotag
00:17 <         thl> | but yes, it could solve this problem, too
00:17 <         thl> | thimm, what's the PC's opinion on the repo tag
these days?
00:17 <       thimm> | Maybe we should decide on repotag, then this is
no issue anymore?
00:17 <       thimm> | I think the decision on using one is ours
00:18 <       thimm> | The PC will tell us how
00:18 <         thl> | thimm, I disagree
00:18 <         thl> | thimm, wehn we want to enforce a repotag then
spec files for EPEL have to look different then those for Fedora
00:18 <       nirik> | so what problem does repo tags solve? easy for
people to see who produced the rpm?
00:18 <       thimm> | No
00:18 <         thl> | thus it's IMHO something the packaging commitee
has to decide
00:18 <       thimm> | specfiles remain the same
00:18 <         thl> | thimm, how?
00:19 <       thimm> | The repotag is part of the %{?dist}
00:19 <         thl> | but dist is optional
00:19 <       nirik> | dist is not required
00:19 <       thimm> | See my packages I moved over from ATrpms
00:19 <       thimm> | nirik: If dist is not used, then that's another
item, and perhaps needs no rebuild (firmware etc)
00:19 <         thl> | thimm, as long as dist isn't enforced this does
not work
00:20 <         thl> | thimm, a repotag makes IMHO only sense if we use
it everywhere
00:20 <       thimm> | Where dist is not used the packages does "manual
disttagging"
00:20 <       nirik> | then non dist packages will have no repo tag either
00:20 <       thimm> | No, no mandatory repotag or mandatory disttag
00:20              * | thl doesn't like what thimm outlines
00:20 <       nirik> | then what does the repotag get us?
00:20 <         thl> | nirik, exactly
00:20 <       thimm> | The repotag discussion is all about extending
%{dist}, nothing more
00:20 <         thl> | thimm, I disagree
00:20 <       thimm> | nirik: see epel-devel, there threads are endless
00:21 <         thl> | a repotag only makes really sense IMHO if it is
used everywhere
00:22 <    mmcgrath> | thl: +1
00:22 <       thimm> | Hm, it does look like all EPEL packages are using
%{dist}, at least at the first glance
00:22 <         thl> | thimm, doesn#t count IMHO
00:22 <         thl> | there are many Fedora pacakges that don#t use it
00:23 <       nirik> | epel-release doesn't IIRC...
00:23              * | thl takes a strong look at the "'#" key -- do
what I mean
00:23 <       thimm> | We don't want to suggest repotags for Fedora
00:24 <         thl> | my opinion: a) enforce a %{repotag} everywhere
(including fedora, even if the repotag is not extended) or b) no repotag
for epel
00:24 <       nirik> | well, do we want to do a wiki based vote on it
then? or ?
00:25 <         thl> | nirik, I'd say we should discuss this on the list
first
00:25 <       thimm> | thl: we can't decide on what Fedora doeas
00:25 <         thl> | to make sure people vote about the same thing
00:25 <       nirik> | it's been discussed a lot on the list already. ;(
00:25 <       thimm> | indeed
00:25 <         thl> | thimm, that's why this is IMHO something for the PC
00:25 <       thimm> | ?
00:25 <       thimm> | The PC is just a technical committee
00:25 <       thimm> | Not political
00:26 <         thl> | nirik, well, seems even here people have
different expectation about the repotag and it's use
00:26 <       thimm> | And the repotag stuff is 99% political
00:26              * | nirik is leaning toward no repotag... I don't
think it gets us that much, and it won't be on everything, so it would
be inconsistant.
00:26 <         thl> | so we IMHO need to agree how to actually do it
before we can vote
00:26 <     entr0py> | nirik: +1, thl +1
00:26 <       thimm> | nirik: It will get us war or peace, I prefer peace
00:26              * | thl is currently leaning toward no repotag, too
00:26 <       nirik> | thimm: who will it get us war with? and why?
00:27 <       thimm> | Centos, Dag, rpmforge
00:27 <       nirik> | perhaps we should look at it from the political
angle...
00:27 <         thl> | thimm, Centos?
00:27 <         thl> | thimm, never heard of that; pointers?
00:27 <    mmcgrath> | thimm: Our concern with the repo tag is that we
don't use dist tags everywhere.  Maybe we should shift the focus to no
repo tag or manditory dist tags.
00:27 <       thimm> | No, no mandatory disttags
00:27 <       thimm> | That doesn't exist in Fedora
00:27 <         thl> | then no repotags
00:27 <       thimm> | And we should not enforce it suddenly out of EPEL
00:28 <    mmcgrath> | Thats the point, why don't we try to make them
manditory (take it to packaging or Fesco)
00:28 <    mmcgrath> | if its that big of a concern.
00:28 <       nirik> | why does centos/dag/rf care about us using
disttags? I guess I should go back and look at the thread on the mailing
list... I can't recall.
00:28 <       thimm> | Personally I'm very much in favour of mandatory
disttags
00:28 <       thimm> | Just not out of this context
00:28 <       thimm> | E.g. mandatory disttags should stand per se and
not be intorduced for repotags
00:29 <       thimm> | But you'll get my vote on mandatory disttags any
day of the week.
00:29 <         thl> | not the mass rebuild with chaning disttags
discussion again
00:29              * | mmcgrath feels this discussion is becoming more
academic then anything.
00:29 <       thimm> | Then let's get pragmatic
00:29 <       nirik> | dist doesn't make sense in some cases where you
don't want to have to rebuild for a new release, but ok...
00:29 <       thimm> | Let's just vote on repotag or not
00:30 <       thimm> | nirik: only whereit does make snes, of course
00:30 <         che> | how about having  a single sheet with pro and con
arguments written down
00:30 <       thimm> | Not firmwar for example
00:30 <         che> | this would make the decision transparent
00:30 <       thimm> | che: care to setup a page?
00:30 <         thl> | che, you know, world isn#t black and white ;-)
00:31 <         che> | thimm, this is how i always did make decisions in
a professional way when i was still working for a process consulting
company
00:31 <         thl> | my vote currently is: repotag only if we have
them in all packages
00:31 <         che> | thl, i live in the perfect world... forgot that?
00:31 <       nirik> | I am thinking no disttags based on tech
arguments, but on the political side I would really like rf/dag to be
happy and join us.
00:31 <         thl> | that's not easily possible right now, so my vote
is no repotags (for now)
00:32              * | thl wonders what drugs che takes ;-)
00:32 <       quaid> | ... and where to get some
00:33 <         che> | hah you wish!
00:33 <         thl> | deadlock?
00:33 <       thimm> | thl: will you prepare the votes for repotag and
rebuilds in the wiki?
00:33 <       nirik> | part of the problem is that the thread between
dag and mschwent really went academic... didn't help to clarify anything
for me.
00:33 <         thl> | for rebuilds yes, as a example
00:33 <         thl> | for repotags: no
00:33            --> | lutter (David Lutterkort)  has joined
#fedora-meeting
00:33 <         che> | thl, well how else could one reproduce how
decisions were made?
00:34 <       thimm> | thl: why not?
00:34 <         che> | thl, how else does one see if there are new
arguments that can fire up another "decision making" process
00:34 <       nirik> | perhaps che would be willing to make a repotags
page for voting?
00:34 <         thl> | thimm, why don't you do it?
00:34 <         thl> | thimm, you seem to be interested in it
00:34 <         che> | nirik, if you are going to work on other things i
have on my todo list sure.
00:34 <         thl> | but such a controversial issue IMHO needs to be
discused again to make sure people vote about the same thing
00:34 <       thimm> | Ok, finish up your vote proposal and I'l copy and
paste ... ... ...
00:35 <         thl> | thimm, no
00:35 <         thl> | you IMHO need to go to the list again
00:35 <       nirik> | well, it could be 'repotags: yes or no' If 'no'
then stop. If yes then 'how needs more discussion'
00:35 <       thimm> | thl: Did chec give you the drugs already?
00:35 <       thimm> | Why can't I write down a vote propoasl?
00:35 <       nirik> | che: :) Wish I had time for the things on my own
todo list. ;)
00:36 <       thimm> | About somethign we discussed weeks ago on the list
00:36 <       thimm> | in two endless threads
00:36 <         che> | nirik, well i feel with you :)
00:36 <         thl> | thimm, nobody stops you
00:36 <         che> | nirik, for me theres a difference between
rational decision and an election
00:36 <         thl> | but I'll oppose a vote where I got the impression
that people don't know the details to actually come to a decisions
00:36 <         thl> | that's importatn in my eyes
00:37 <       nirik> | how about someone who wants dist tags writes a
short (small paragraph) PRO: section for the wiki vote, and someone
against writes a CON: (like the do for real elections here)
00:37 <         thl> | details in this case: %{?dist} or %{?repotag}
00:37 <       thimm> | thl: That's a bad way to go
00:37 <         thl> | use it everyhwere or now
00:37 <       thimm> | That way everyone not liking a vote can "oppose"
to it
00:37 <         thl> | thimm, it worked fine for FESCo afaics, as it
works like this
00:38 <         che> | nirik, i am just sad that i cant stress this
topic without going offtopic :)
00:38 <       thimm> | Anyway, thl, anyone around here can throw in
something for voting, right?
00:38 <       thimm> | Let's move on
00:38 <       nirik> | thimm: I would think so. If people fear they
don't have the info they need they can say no or abstain.
00:39 <       thimm> | nirik++
00:39 <         thl> | before we move on one thing
00:39 <         thl> | nirik, thimm, others: what do you prefer: delete
and rebuild or add a .1 and rebuild?
00:39 <       nirik> | delete and rebuild is what I would prefer.
00:39 <       thimm> | the former
00:39 <         thl> | nirik, well, agreed in parts, but it might look
bad to the outside
00:40 <       nirik> | anyone using our not announced rh5 rpms that were
compiled against beta should know better...
00:40 <         thl> | nirik, you know that some people out there will
have different packages under the same name?
00:40 <         thl> | the debuginfo pacakges won#t match
00:40 <       nirik> | what do you mean? same name?
00:40 <         thl> | and in bug reports you don#t know which of the
pacakges people are using
00:40 <         thl> | nirik, say I isntall a package from EPEL5 today
00:40 <         thl> | nirik, we delete and rebuild
00:40 <       thimm> | but you assume actual users of the pre-repo
00:41 <         thl> | nirik, user has a problem in two weeks from now
00:41 <         thl> | nirik, debuginfo package does not match
00:41 <         thl> | nirik, bug refers to pacakge foo-1.1
00:41 <       nirik> | sure. So in the bug report: please remove and
instlal the new one.
00:41 <         thl> | and people don#t know if that the old or the new one
00:41 <       thimm> | thl: how die the user get the package in the
first place?
00:41 <         thl> | thimm, repo is public since some weeks now
00:41 <       nirik> | I don't think it will be that much of a problem.
00:41 <       thimm> | s/die/did/
00:42 <       thimm> | Yes, public means not that one randomly installs
stuff, right=?
00:42 <       thimm> | It isn't installed
00:42 <       thimm> | s/installed/announced/
00:42 <         thl> | thimm, well, some people might have installed
stuff from it already
00:42 <       thimm> | Some peopl may have installed from SLES, too
00:42 <    mmcgrath> | Some people have installed stuff, and announced
stuff on their own already.
00:43 <         thl> | adding a .1 isn't such a big deal IMHO
00:43 <       thimm> | mmcgrath: which ones?
00:43 <         thl> | well, let's move on
00:43              * | thl waits
00:43 <    mmcgrath> | thimm: just random people on the net.  I don't
think anything major.
00:45              * | thl will move on
00:45            --- | thl has changed the topic to: EPEL Meeting
00:45 <         thl> | so, what else?
00:45 <     entr0py> | any comments on my request for
brp-python-bytecompile?
00:45 <         thl> | quaid, there is "Investigate single RHEL
subscription for EPEL maintainers" on the schedule
00:46 <       thimm> | entr0py: brp-python-bytecompile is currently
under discussion for other issues, too
00:46 <       thimm> | So an improved version may appear in *-rpm-config
00:46 <       thimm> | Which would fix your issue as well
00:46 <         thl> | thimm, will the PC take care of the bytecompile
stuff?
00:47 <     entr0py> | thimm: excellent
00:47 <    mmcgrath> | thl: I don't think that would happen any time soon.
00:47 <       thimm> | About thy pyc/pyo stuff
00:47 <    mmcgrath> | not to discourage but it would be a long time
before we could actually have something.
00:47 <       thimm> | But at the end it will be a suggestion for the
redhat-rpm-config maintainer
00:47 <         thl> | mmcgrath, the idea wasn't mine; I actually agree
with you
00:48 <       thimm> | mmcgrath: Yes, this will not be done soon
00:48 <       nirik> | so voting on the wiki would take place until
when? next meeting?
00:48 <         thl> | nirik, I'll annouce it on the list
00:48 <       nirik> | ok.
00:48 <         thl> | anything else?
00:49              * | nirik looks at the schedule page
00:50 <    mmcgrath> | OH,
00:50 <    mmcgrath> | I've got one thing.  If its been discussed sorry
I missed it.
00:50 <    mmcgrath> | So RHEL5's $RELEASEVER translates to 5Server
instead of 5
00:50 <    mmcgrath> | Should we have the mirrors list script 'correct'
this?
00:51 <       thimm> | You mean making this a "5"? +1
00:51 <       thimm> | cobbler also bails out on 5Server
00:51 <         thl> | mmcgrath, could we have a symlink maybe that
fixes this?
00:51 <       thimm> | But, it's an easyfix
00:51 <    mmcgrath> | I'm not sure how well the mirrors handle symlinks.
00:52 <    mmcgrath> | I'd think the mirrors can handle symlinks fine
but I don't want to assume that.
00:52 <         thl> | mmcgrath, so we should hardcode it in the repo
files?
00:52 <    mmcgrath> | We can, thats what I've done on my boxes.
00:53 <         thl> | hardcoding seems fine to me
00:53 <       thimm> | +1
00:53            --> | sharkcz_ (Dan Horak)  has joined #fedora-meeting
00:53 <    mmcgrath> | k, I won't worry about it then :)
00:54 <         thl> | so I'll tell stahma to update the repo files to
use a hardcoded 5 instead of $RELEASEVER
00:54 <         thl> | that okay for everybody?
00:54              * | thl waits a bit if someone yells
00:54 <       nirik> | thl: +1
00:55 <         thl> | seems nobody yelled
00:55 <         thl> | anything else?
00:55              * | thl will close the meeting in 60
00:55 <    mmcgrath> | not here
00:55 <         thl> | btw, any volunteers for writing the summary?
00:56              * | thl will close the meeting in 30
00:56              * | thl will close the meeting in 10
00:56 <         thl> | -- MARK -- Meeting end
}}}




More information about the epel-devel-list mailing list