Ubuntu Wiki

Log

TZ UTC-5

(07:00:46 AM) MootBot: Meeting started at 12:00. The chair is dholbach.
(07:00:46 AM) MootBot: Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE]
(07:00:54 AM) TheMuso: persia: You're welcome to do minutes if you want
(07:00:54 AM) dholbach: Welcome to another MOTU meeting
(07:01:11 AM) dholbach: our Agenda is available at: https://wiki.ubuntu.com/MOTU/Meetings
(07:01:28 AM) dholbach: first up is persia
(07:01:36 AM) dholbach: [TOPIC] Discussion of change from requesting interdiff files to requesting diff.gz files. Despite documentation, interdiff seems hard to use.
(07:01:36 AM) MootBot: New Topic:  Discussion of change from requesting interdiff files to requesting diff.gz files. Despite documentation, interdiff seems hard to use.
(07:01:38 AM) persia: TheMuso: OK.  If you don't mind announcements: I tend to forget :)
(07:01:50 AM) TheMuso: persia: Announcements are fine.
(07:02:08 AM) persia: Interdiff files are wonderful, do everything we need, and help with the review process.  Unfortunately, despite documentation, many people seem to have trouble with them.
(07:02:59 AM) persia: cjwatson pointed out that most of the interesting information is contained in the diff.gz, and that the diff.gz was actually smaller than the "full interdiff".  In testing, I've found that I can reliably produce a .changes file for a new upstream from a diff.gz.
(07:03:22 AM) dholbach: I personally haven't dealt much with interdiffs yet, but found it OK once I knew about the process-interdiff script.
(07:03:24 AM) TheMuso: But interdiff files are a nice way to see exactly what has changed.
(07:03:30 AM) persia: Does anyone know of any reasons not to shift from interdiff to diff.gz in the hopes of making the process easier to understand?
(07:03:55 AM) persia: TheMuso: True, but the understanding is easier with -p1, and that makes it hard to reconstruct the target package.
(07:03:57 AM) TheMuso: Yes, looking at .diff.gz alone can sometimes be hard to work out what has changed.
(07:04:28 AM) TheMuso: persia: Following the documented process at http://wiki.ubuntu.com/UbuntuDevelopment/Interdiff, I have had no problems making interdiffs for others, and reviewing them.
(07:04:38 AM) persia: Most of the confusion seems to be related to the presence or absence of -p1.  By migrating to diff.gz, we'd push the burden of understanding when to use -p1 on the sponsor.
(07:04:58 AM) TheMuso: Right.
(07:05:03 AM) persia: TheMuso: Neither I, but I've seen others with difficulties.
(07:05:24 AM) TheMuso: What is so difficult about following a process?
(07:05:30 AM) TheMuso: That is documented/
(07:05:34 AM) ***persia doesn't know
(07:05:54 AM) TheMuso: I fear that the few of us who are present, who seem to be fine ith them cannot make this decision.
(07:06:05 AM) TheMuso: We need more attendees, particularly those who have had trouble with them, to speak up.
(07:06:21 AM) TheMuso: s/speak up/attend the meeting and put their view forward/
(07:06:49 AM) dholbach: I feel the process decision would benefit from an evaluation period and the script being more public (ubuntu-dev-tools maybe) - maybe also a wiki page at PackagingGuide/Recipes
(07:07:07 AM) TheMuso: dholbach: Agreed. I was not even aware that the script was in a state that was usable.
(07:07:14 AM) TheMuso: i've been doing it manually myself.
(07:07:35 AM) persia: dholbach: It's a one-line call, documented at both UbuntuDevelopment/Interdiff and MOTU/Contributing.  The script isn't very reliable, although RainCT is looking at it.
(07:07:42 AM) dholbach: just to avoid going through the "ok, we've changed the process again", "hey we reverted it because somebody complained", etc business
(07:08:05 AM) dholbach: persia: I think it'd help to distribute it and improve it in iterations of using it day-to-day
(07:08:22 AM) dholbach: I'm sure we can tweak it to a point where it makes people happier :)
(07:08:30 AM) dholbach: ... even happier... :)
(07:08:34 AM) persia: Sure.  I don't see any point to changing the process unless one of the people who complained has something to say in the meeting.  Two weeks ago, this was raised in "Any Other Business", and all comments were positive, but I promised to investigate alternatives, hence this topic.
(07:09:34 AM) dholbach: Ok... what can we agree on and what can we transform into action items?
(07:09:41 AM) ***siretart is late - sorry
(07:10:24 AM) dholbach: somebody who wants to help me with setting up a recipe page (in the new year)?
(07:10:29 AM) persia: In the absence of discussion from anyone with problems with Interdiff, let's agree that we should make more efforts to get automation scripts working well.
(07:10:41 AM) TheMuso: I agree with that.
(07:10:55 AM) siretart: I tried interdiff for one of my packages
(07:10:59 AM) dholbach: persia: will you add your process-interdiff script to ubuntu-dev-tools as a first step?
(07:11:13 AM) persia: dholbach: Not in the current state.  It breaks too often.
(07:11:31 AM) cjwatson: why were we ever requesting people to use interdiff by hand rather than debdiff?
(07:11:37 AM) cjwatson: (debdiff uses interdiff internally)
(07:11:58 AM) persia: cjwatson: new upstream packages sometimes have new binary objects which debdiff can't represent.
(07:12:02 AM) siretart: cjwatson: I assume because debdiff also displays the upstream changes
(07:12:14 AM) siretart: cjwatson: perhaps we can tune debdiff to skip those?
(07:12:59 AM) siretart: persia: this sounds like a bug to me. I'd expect that it notes this fact, but continues then
(07:13:22 AM) persia: siretart: Yes, but it means that one can't reconstruct the target package from the debdiff.
(07:13:55 AM) persia: One can reconstruct from interdiff -p0 or diff.gz from the new package.  There may be other ways, which I have yet to discover.
(07:14:03 AM) siretart: persia: is that really necessary?
(07:14:44 AM) persia: siretart: I think so.  I'd rather have a relatively small single file to download as a bug attachment when sponsoring a new upstream than negotiating from where to get a package.
(07:15:05 AM) persia: (especially when sponsoring from unknown contributors)
(07:15:14 AM) siretart: hm. I see
(07:15:30 AM) ***Hobbsee notes that those who are not up with the latest motu proceedures still use debdiff, quite happily
(07:15:31 AM) siretart: how about teaching debdiff binary diffs then? (perhaps using xdelta or something)
(07:15:49 AM) ***Hobbsee notes that you're going to get resistance switching everyone over, for something that seems to be of little benefit
(07:16:11 AM) persia: siretart: Considering that we can usually just download orig.tar.gz from upstream, and that we should when we don't know or trust the contributor, I'm not sure that's ideal.
(07:16:43 AM) dholbach: cjwatson: what are your thoughts about this?
(07:16:48 AM) persia: Hobbsee: Sure.  Maybe takes a while.  Still, good to pick the right thing before pushing everyone to switch.  Interdiff has been documented in one way or another for about 6 months.
(07:17:27 AM) ***norsetto thinks that downloading a diff.gz from a bug report and then a tarball from somewhere is not much different than downloading a diff.gz and a tarball from revu
(07:17:44 AM) ***siretart agrees to norsetto 
(07:18:13 AM) ***persia thinks that new upstreams have a lower threshold for changes than REVU packages, and that this is confusing to Contributors.
(07:18:36 AM) ***persia further doesn't trust tarballs on REVU, having found a few that were different for no clear reason
(07:19:02 AM) cjwatson: dholbach: personally? not being able to open a patch directly in my browser makes me less likely to review it in a timely fashion
(07:19:17 AM) TheMuso: cjwatson: That I can understand.
(07:19:19 AM) cjwatson: I think we should fix the tools rather than reverting to a process that will delay review
(07:19:43 AM) persia: cjwatson: Do you mean adding hunking support to combinediff?
(07:19:43 AM) cjwatson: and work around it on a case-by-case basis in the odd instance where there's a problem
(07:20:03 AM) cjwatson: persia: I do think new upstreams are a special case
(07:20:15 AM) cjwatson: and there's no reason to throw the non-new-upstream baby out with the bathwater
(07:20:25 AM) persia: cjwatson: Ah.  Currently Interdiffs are only recommended for new upstreams.  All other patches are supposed to be debdiffs.
(07:20:32 AM) cjwatson: oh, I see, I misunderstood then
(07:20:43 AM) persia: We're trying to close the corner case.
(07:20:55 AM) cjwatson: yes, I think in that case a full .diff.gz is probably easier to review
(07:21:00 AM) cjwatson: so I take back my comments
(07:21:41 AM) cjwatson: but new upstream bugs should definitely come with clear instructions on retrieving the .orig.tar.gz
(07:22:21 AM) persia: I argue those instructions belong in either a watch file or a get-orig-source target in debian/rules
(07:23:13 AM) cjwatson: I agree that they should be there, but relying solely on them complicates review
(07:23:22 AM) cjwatson: and it means that you have to run stuff from the package before you can review it
(07:24:07 AM) persia: cjwatson: That's the reason for writing the automation scripts to pull it out.  Good point about running unseen code in get-orig-source though.
(07:24:43 AM) cjwatson: it doesn't seem much to ask for somebody to provide a URL in addition
(07:25:16 AM) geser: persia: I haven't looked at it yet: but how does it work to call the get-orig-source target from the diff.gz? how to you apply the diff.gz without the source tarball?
(07:25:25 AM) persia: geser: filterdiff :)
(07:26:12 AM) dholbach: to me it seems we should hold off on making a final process decision today but work on documentation and wiki examples instead and give people some time to evaluate the process
(07:27:12 AM) persia: That seems reasonable to me, although with Interdiff having been documented for the last six months, I'd like to continue recommending that to new Contributors until and unless we pick another process (perhaps using diff.gz).
(07:28:04 AM) dholbach: I didn't intend us to revert any decision or documentation now.
(07:28:17 AM) persia: OK.  That sounds perfect then :)
(07:28:20 AM) dholbach: great
(07:28:24 AM) TheMuso: +1
(07:28:33 AM) dholbach: persia: shall we work together on documentation and recipes?
(07:29:07 AM) persia: dholbach: I don't really understand recipes as implemented on the Wiki.  I'm happy to write yet more docs if the current are insufficiently clear.
(07:30:02 AM) dholbach: persia: the recipes are intended to be step-by-step examples, so people would download a tarball, an existing package and the steps would guide them to a working interdiff etc
(07:30:31 AM) persia: dholbach: Right.  We can discuss in another forum.  Consider my previous response mostly as "yes".
(07:30:38 AM) dholbach: [AGREED] No process change now, improve documentation (examples) instead and give us some time get feedback on it.
(07:30:38 AM) MootBot: AGREED received:  No process change now, improve documentation (examples) instead and give us some time get feedback on it.
(07:30:58 AM) dholbach: [ACTION] persia and dholbach to work on wiki documentation containing examples
(07:30:58 AM) MootBot: ACTION received:  persia and dholbach to work on wiki documentation containing examples
(07:31:00 AM) dholbach: thanks a lot persia
(07:31:08 AM) dholbach: [TOPIC] Other business
(07:31:08 AM) MootBot: New Topic:  Other business
(07:31:26 AM) dholbach: any other business?
(07:31:38 AM) dholbach: going once...
(07:31:43 AM) TheMuso: Should we perhaps notify people that there may be few/no MOTUs around for the next two mondays, due to the break?
(07:32:02 AM) TheMuso: and people doing the festive things they do?
(07:32:19 AM) dholbach: TheMuso: I believe persia already sent out a notice
(07:32:25 AM) TheMuso: Oh?
(07:32:27 AM) persia: TheMuso: I've recently sent an annoucement about REVU days, but something more general may be useful.
(07:32:32 AM) TheMuso: persia: Ok.
(07:33:02 AM) dholbach: Does somebody volunteer to send out a more general announcement?
(07:33:33 AM) TheMuso: persia: Where did you send that announcement? I don't think I've seen  anything to that effect yet.
(07:33:41 AM) norsetto: why general if its about mondays? Or is that about the next two weeks?
(07:33:42 AM) dholbach: TheMuso: -motu and -motu-mentors
(07:33:45 AM) persia: TheMuso: About 5 minutes before the meeting started.
(07:33:51 AM) TheMuso: persia: ah ok
(07:34:09 AM) dholbach: norsetto: yeah, the next two weeks - I expect less activity
(07:34:13 AM) persia: norsetto: About the next two weeks (for general MOTU availability).  That wasn't clear in my mail.
(07:34:13 AM) TheMuso: Hrm. Mail is slow then.
(07:34:14 AM) dholbach: (at least speaking for myself... :))
(07:34:38 AM) norsetto: dholbach: you bugger :-)
(07:34:56 AM) dholbach: Ok... do we need a more general announcement?
(07:35:02 AM) geser: TheMuso: https://lists.ubuntu.com/archives/ubuntu-motu/2007-December/002942.html
(07:35:05 AM) dholbach: maybe I'll just do it
(07:35:15 AM) TheMuso: geser: thanks
(07:35:19 AM) dholbach: [ACTION] dholbach to send out a general announcement about christmas activities
(07:35:20 AM) MootBot: ACTION received:  dholbach to send out a general announcement about christmas activities
(07:35:23 AM) dholbach: any other business?
(07:35:36 AM) dholbach: going twice...
(07:35:49 AM) dholbach: ok, gone
(07:35:50 AM) dholbach: [TOPIC] date and time of next meeting
(07:35:51 AM) MootBot: New Topic:  date and time of next meeting
(07:36:01 AM) ***persia proposes 20:00 UTC 4th January
(07:36:28 AM) TheMuso: Good for me.
(07:36:49 AM) ***dholbach keeps quiet because he'll likely be on the road then
(07:37:15 AM) geser: good for me too
(07:37:21 AM) dholbach: ok... let's go with 20:00 UTC 4th January then
(07:37:30 AM) dholbach: [AGREED] next MOTU meeting is 20:00 UTC 4th January
(07:37:31 AM) MootBot: AGREED received:  next MOTU meeting is 20:00 UTC 4th January
(07:37:52 AM) TheMuso: I'll send an announcement out as soon as I return next thursday.
(07:37:53 AM) dholbach: adjourned - thanks for attending the meeting :)
(07:38:07 AM) dholbach: thanks TheMuso and persia for sending out those mails
(07:38:13 AM) dholbach: #endmeeting
(07:38:13 AM) MootBot: Meeting finished at 12:38.