IRC log for #koha, 2007-03-16

← Previous day | Today | Next day → | Search | Index

All times shown according to UTC.

Time Nick Message
11:04 kados paul_away: you still around?
11:05 hdl: can you call paul and find out if he has a copy of 2.2.8 we can put on koha.org?
11:05 hdl: /me doesn't have one
11:05 mail to koha-devel sent
11:07 hdl found one
11:09 Do you want me to send it to you ?
11:10 or do you want to get  it on a website ?
11:18 kados ?
11:18 dewey hmmm... kados is helping us LibLime folk with something at the moment
11:19 kados hdl: sorry
11:19 hdl: phone call
11:19 hdl: can you email it to me?
11:19 hdl: jmf@liblime.com?
11:19 hdl: maybe put it on a website too?
11:20 hdl: :-)
11:20 hdl ok.
11:31 on koha-fr.org
11:31 koha-2.2.8 disponible
11:32 kados thanks!
11:32 where?
11:33 hdl: i don't see a news item, or a download link
11:34 hdl http://www.koha-fr.org/article.php3?id_article=93
11:34 (now in recent articles)
12:02 kados hdl: I've changed koha.org links, but they must be 'approved' first by russel
12:21 owen kados, did I see that you had made some kind of change in rel_3_0 that allowed for custom headers in templates?
12:22 Or rch, maybe you know?
12:27 Hi slef
12:27 slef hi
12:27 dewey bonjour, slef
12:29 slef please not google and SVN?  A double-whammy of bad technology, just because some people won't do local work in a DVC and you don't want to wait one more day for the service to return?
12:30 owen I take it you don't like SVN, slef?
12:31 slef indeed... it's replacing one buggy legacy tool (CVS) with a much larger buggy legacy tool
12:31 If you're going to suffer the pain to fix it, do it properly.
12:31 kados hdl: owen that was me
12:32 owen: yea, I basically just took the title tab out of the doc-head
12:33 slef: all we have is 'some people' :-)
12:33 slef kados: huh?
12:34 kados IMO DVCs create unnecessary overhead in managing version control, and in a small community, we can't really afford it
12:35 plus, we have a hard enough time getting people to wrap their minds around CVS ...
12:35 if we switched to something like git or arch, we'd never get any user contribs
12:36 slef I can't see how you work that out.  The release manager has to publish their repo, but they do that with a centralised service.
12:36 Meanwhile, they save time by being able to work locally when needed, only doing the expensive publish when necessary.
12:36 kados we can do the same thing in SVN with branches
12:37 since we only have 10 or less active developers, that should scale fine for now
12:37 slef I'm not going to argue with your guesses about the future.  I just disagree with them.
12:37 kados sure, and I certainly appreciate your points
12:37 slef CVS scales this far, so why the big pain and bigger clients of SVN?
12:38 kados for what it's worth, hdl agrees with you
12:38 the main reason is that CVS hasn't scaled for us
12:38 we've run into barriers with CVS that SVN solves
12:38 slef such as?
12:39 kados seemingly simple things like being able to rename directories, files
12:39 and changing permissions
12:39 merging between branches
12:39 are a real pain in CVS
12:40 slef They're as much a pain in SVN.  Still need to read the manual.
12:40 Renames excepted.
12:40 kados well we've been using svn at liblime
12:40 so it's not like I'm just making a snap judgement
12:40 we actually really like it
12:41 slef how many of the other candidates have you used?
12:41 kados well, arch I've used
12:41 chris has used git
12:41 slef we use git for pretty much everything at ttllp
12:41 kados I haven't found anything that I want to do that I can't in SVN
12:41 slef including stuff like koha where we talk to CVS outside
12:42 kados I really don't want to spend much bandwidth on the version control system
12:42 rather devote energy on actual development
12:42 slef So you don't want consensus?  You just want to lumber us with SVN@Google and that's it.  End of discussion.
12:43 kados lumber?
12:43 slef saddle, curse, weight down, ...
12:43 kados hehe
12:43 yea, I know what it means
12:43 I just don't see how it's going to do that
12:44 it's fast, got plenty of backing from a strong organization, with core developers on staff
12:44 slef Why Google?  Google is a large evil corporation, rather contraversial in the library community and has just been accused of massive copyright infringement.
12:44 kados Google is a large evil corporation?
12:45 are you kidding?
12:45 slef No.
12:45 kados honestly, I don't have time for this
12:45 slef When will you?
12:46 kados I'd be happy to respond to an email you send to the list
12:46 but I'm not going to battle it out with you on IRC when you're citing a reason for not going with google is be cause 'google is evil
12:47 that's an opinion, and shouldn't be stated as fact
12:47 IMO :-)
12:47 slef My comments on why google is evil have been public for at least a year.
12:51 owen kados: I've been playing around with YUI's DataTable library, trying to attach it to a copy of an OPAC screen.
12:51 It's working pretty well, with a few snags.
12:51 kados sweet
12:51 owen But I'm wondering about strategies for including the required scripts
12:52 That's why I was curious about changes to the templates
12:52 kados well, you have two options: host it at yahoo, or store it locally
12:52 I think if we coded carefully, we could make that a sys pref
12:52 owen Right, but for each page that has a data table you'd need to supply a snippet of custom javascript defining the parameters of that table
12:53 kados yea, the changes basically just put the <title>page title</title> bit entirely in the .tmpl file
12:53 rather than in the doc-head incldues
12:53 so you can add new doc-head includes on a template by template basis
12:53 does that make sense?
12:54 owen I'm not sure... I'm also not sure about how other applications handle per-page javascript stuff. Do they include it in the scripts that get loaded for the whole site? That seems cumbersome to me, but I don't really know about the overhead
12:55 I do know that in my local tests using the Yahoo-hosted libraries really slows down page load.
12:55 kados well, that's actually why I did it -- I don't want to load all the yahoo stuff every time
12:55 it's not going to all need to be on every page
12:56 owen Okay, so that's an issue addressed by your changes.
12:56 kados yep
12:57 owen Cool. Good to know. The DataTable library is still beta, and it shows...I've already run across a couple of problematic bugs
12:57 kados have they been reported yet?
12:57 what are you using DataTable for?
12:57 out of curiosity
12:58 owen Making tables sortable.
13:04 kados hehe, I did that too :-)
13:04 it's pretty slick
13:04 that was the first thing I did after I played with grids
13:10 slef: I'll happily respond to an on-list post about why you thing SVN/Google is a bad idea
13:11 slef kados: just writing.
13:13 kados thx
13:26 slef sent btw
13:40 right, I'm afk for a few hours... please don't break the project while I'm out ;-)
14:00 kados slef: responded
14:33 owen-away Am I out of touch? Do librarians really think Google is evil?
14:55 tnb owen: I heard some rumblings a few years back, but nothing lately.  t
14:56 owen I know librarians don't like the fact that Google is replacing them as the go-to guy for information
14:56 tnb yeah, that's true
14:56 owen And I know /publishers/ don't like the fact that Google is scanning books.
14:56 tnb sure
14:56 owen But I don't know why librarians should care about that
14:56 I would think librarians would embrace increased access to texts
14:57 kados yea, to me google's been a real advocate of open access to data
14:57 I mean, the reason they're being sued over youtube is because of that very issue, right?
14:58 tnb plus, they've hired a company chef to cook healthy gourmet food for their employees ;)  
14:58 owen Yeah, the fact that Google is being sued over YouTube seems more of a badge of honor than a flaw
14:58 tnb ha!
14:58 owen :D
14:59 kados--you hit the nail on the head.
15:01 Also, when Google's detractors look like this it's hard to take them seriously: http://www.google-watch.org/
15:01 kados hehe, that's a funny picture of bill
15:02 it actually endears him to me :-)
15:17 hdl: you there?
15:17 hdl: I just got a report that the file is corrupted for 2.2.8
15:18 hdl: confirmed:
15:18 gzip: stdin: unexpected end of file
15:18 tar: Read 7908 bytes from koha-2.2.8.tar.gz
15:18 koha-2.2.8/intranet-html/intranet-tmpl​/default/en/bull/searchresultlist.tmpl
15:18 koha-2.2.8/intranet-html/intranet-tmp​l/default/en/bull/serial-issues.tmpl
15:18 tar: Unexpected EOF in archive
15:18 tar: Unexpected EOF in archive
15:18 tar: Error is not recoverable: exiting now
16:31 hdl kados : SH.
16:31 kados ?
16:33 tnb owen: you around?
16:33 owen Yes
16:34 tnb this if off Koha topic... but I seem to remember us discovering
16:34 that we could do 'user groups' with WordPress (when we were investigating voting site)
16:34 owen (did hdl just shush kados??)
16:34 tnb ha! I don't know...
16:34 hdl No. It was a bad word.
16:34 tnb oh, uh oh :/
16:35 wha happened?
16:35 hdl kados : I found other tarballs.
16:35 tnb kados is on the phone at moment
16:35 owen Heh, hdl try "SH**"
16:35 hdl I hope they are not corrupted either.
16:35 owen tnb, what do you mean by user groups?
16:35 tnb owen: do you remember this user group thing, because I cannot at all find which plugin does it now.  yep
16:36 so a person is only in charge of moderating
16:36 certain user group
16:36 i've been on the wordpress site... was there a hack you rememmber?
16:37 owen I have no recollection of that. Can you describe in more detail what you're trying to do?
16:37 hdl Sorry I didnot want to be rude.
16:37 tnb hdl: no prob :)
16:37 kados still on phone :L)
16:38 owen: we have several profs adn want them to be in charge of moderating just 'their' students posts
16:38 so, a 'user group' for each LIS classroom and when folks sign up, they select their classroom, then prof only responsible for moderating their own students posts (i guess i just repeated myself :/ sorry)
16:38 hdl kados : I send you a new copy I could untar right now.
16:39 tnb do you remember any workaround for soemthing like that
16:39 ?
16:39 owen Sorry tnb, I don't recall working on anything like that. In fact it sounds pretty off-the-wall. I'd be surprised to find something like that pre-written.
16:40 tnb huh.  ok.  it was a longshot :)
16:45 kados hdl: hi
16:45 hdl: so you have a fixed version?
16:45 hdl kados : I just sent you one. (There were 3 versions on paul's machine.)
16:46 kados hdl++
16:46 hdl Luckily, I came to my computer at nearly 9PM
16:47 kados hdl: thanks!
16:47 hdl: it's live now
16:47 hdl You're welcome
16:49 kados hdl: I tested it, it unpacks successfully
16:49 hdl: hopefully it's the right 2.2.8 :-)
17:37 slef kados: you are not reading what I write.  Your prejudices are showing.
17:37 'Google is a corporation and therefore evil' is imaginary.
17:38 I'm a member of one of the UK's largest retail corporations.  I don't agree that corporations are necessarily evil.
17:38 And it's ugly that the discussion moves to demonising so quickly :-(
18:02 kados slef: I haven't demonized anyone
18:24 hi _paul
20:16 slef kados: I can bring up a test git+cvs server tomorrow.
20:16 chris heya slef
20:16 kados chris: slef and I have been chatting on #gnu
20:16 slef: chris and I have been chatting on pmest
20:16 slef trouble is, the server I'd have to use is going to die and its replacement isn't ready for production yet
20:16 #savannah
20:16 kados pmesg even
20:17 ahh, right
20:17 chris do you think it would be possible to write up use case of how git and the koha team would work?
20:17 kados slef: http://en.wikipedia.org/wiki/Git_(software)
20:17 Git's design is a synthesis of Torvalds' intimate knowledge of maintaining a large distributed development project, and of file system performance.
20:17 slef sure... can you collect the tasks?  I'd probably be ripping much from 'everyday git for kernel hackers'
20:18 kados slef: A core assumption in Git is that a change will be merged more often than it is written, as it is passed around various reviewers.
20:18 slef git's big problem is that Windows support used to be awkward IIRC
20:18 kados slef: I'm trying to figure out if we're 'a large distributed project' and also if we want to spend more time merging than writing
20:19 slef kados: it's heading that way and almost certainly will if we have 2.2, 3.0s, 3.0z and 3.1 all around at once!
20:19 kados slef: my core fear with DVS and Koha is that with only 11 developers, we won't have any incentive to manage a 'project' repository
20:19 and we'll end up just forking
20:19 and fragmenting into 3-4 separate projects
20:20 slef I think the release managers need to keep the project repo.
20:20 kados yea, but the prob is, we don't have anyone with time to just handle merging
20:20 slef There's always an incentive for others to pull from the project repo: get into the next release.
20:20 kados that's my core intention
20:20 s/intention/contention/
20:21 slef: can you quell my fear regarding that?
20:21 slef We could maybe try to use a robot that tries to merge anything approved.
20:21 kados what's the process of approval consist of?
20:22 slef Most release managers I know bother to check what's been contributed.
20:22 chris its what we really should be doing now anyway
20:22 *snap*
20:22 slef Whatever we want.  GPG-signed, from a particular address or whatever.
20:22 chris hmm that could work
20:23 approved developers get merged automagically
20:23 slef I've done that before, but not with git.
20:23 kados I'm skeptical that we have resources to :
20:23 1. set that up
20:23 2. maintain it
20:23 also, we're at a very important time in the project's hisotry
20:24 we really need to roll 3.0 out
20:24 slef Our git-use has been mostly ssh-based with few people.  Koha's a bit bigger, but given git scales to linux, I doubt there'll be an impossible problem.
20:24 kados we're competing with evergreen now
20:24 3.0 is literaly years behind schedule
20:24 slef I can understand the scepticism (I've been crap before) and can only ask for time.
20:25 chris one issue that we have now, is that we are really just avoiding the problem
20:25 slef I've been hampered with 3.0 work partly because each branch switch feels like a big networked operation.  I guess I could keep two working copies.
20:25 chris ie not reading the commits, not rejecting bad ones etc
20:26 kados google + SVN means I can focus on coding and won't have to worry about the versioning system
20:26 chris: yea, that's true
20:26 slef I've also struggled with 3.0 because it's undocumented, even in code comments.  Even the code style was random last time I looked.
20:26 kados slef: amen
20:26 slef: the search api's well documented
20:27 slef: but that's all that I've seen that has anything remotely like code documentation
20:27 slef IMO the problem is that too many focus on code and no-one is focusing on integration, but maybe that's wrong.
20:27 chris no i think there is more than a grain of truth in that slef
20:27 slef A lot seems to go on on IRC and not be written elsewhere, so my impressions these days are probably a bit off.
20:27 kados slef: yea, both good points
20:28 our core problem is resource related
20:28 we're all just scraping to get the stuff done for our clients
20:28 don't really have time for the project proper
20:28 chris we let stuff slip, and it ends up costing us in the long run, but we need to eat in the short run
20:28 kados yea
20:28 it's a viscious cycle :(
20:28 slef vicious
20:29 kados yea
20:29 chris or viscous (spelling?)
20:29 the sticky stuff
20:29 kados hehe
20:29 slef chris: a viscous cycle is a bike that has just been oiled.
20:29 chris :0
20:30 kados probably all we need
20:30 slef My problems with just switching to google + SVN is it doesn't address either of the main problems with savannah + CVS.
20:31 kados phone
20:31 slef I see the main problems as failure-resilience and centralisation.
20:31 kados hehe
20:31 slef ok, someone's just asked me stuff in #savannah
20:31 s/centralisation/branching/ # probably clearer what I mean
20:31 kados I think the main problem we have is there's no-one handling merges
20:31 chris yeah
20:32 kados the thing we're annoyed about EG for having :-)
20:32 chris maybe git would force us to
20:32 at citylink they have cone man
20:33 someone on call that day (with the traffic cone)
20:33 maybe we need merge person (to be all non gender biased)
20:33 :)
20:33 kados hehe
20:33 chris a role rather than a person, that rotates
20:33 so feel free to say "thats retarded"
20:34 kados sorry, russ is distracting me :-)
20:34 I think it's a good idea
20:34 slef no, but how does that person get paid? ;-)
20:34 kados what kind of person
20:34 would be good at that?
20:34 and who in our community is that kind of person?
20:34 and how will they get paid? :-)
20:35 slef to be honest, stuff I currently do to koha doesn't get merged back promptly at the moment because there seems little incentive to do so
20:36 (and I'm often not developing it on a machine with a cvs-friendly network connection - or my ssh keys for that matter)
20:36 chris right
20:37 slef on the flip side, I like reading code commit mails... I'm still on commit mailing lists for projects I've not used for over a year
20:37 chris yep
20:38 slef I quite liked compiling the 'what changed in koha this week' mails, but no-one else seemed to appreciate them.
20:38 chris ohh i liked those
20:38 sorry i didnt verbalise that
20:39 slef chris: we're not usually awake at the same time, so I don't hold it against you.
20:39 chris i think there are a few ways to be paid for the role
20:39 slef It also got a bit tricky with the split branches.
20:40 chris but mostly it would boil down to the persons employer being willing to pay them for some time to do it
20:41 also they may be willing to do some work for non monetary gain outside work hours
20:41 for things like kudos :)
20:41 i certainly wouldnt mind having a go at it, but certainly not on anything like a fulltime basis
20:42 slef Heh, I'm my employer.  I don't have enough koha clients at the moment to justify the time I spend :-/  Even oscommerce is more profitable.  I stick around because for non-monetary reasons.
20:42 chris but bringing it back .. i think this a problem we will have with whatever version system we use
20:42 the resource to audit code commits
20:44 kados chris: so I take it you prefer not to be the 'auditor' for Koha commits?
20:45 chris no thats fine, i dont mind doing that
20:45 just not forever
20:45 and not if its gonna make my employer go broke :)
20:46 kados hehe
20:46 chris if i was doing it, id probably ask for help occasionally too
20:47 as there undoubtedly will be code i dont grok
20:47 slef thing is, do we need an auditor, or someone saying "this can't be committed because I don't understand it"?
20:47 chris but at least id be going "huh???"
20:47 kados slef: yea, maybe
20:47 chris and asking ppl to explain it
20:48 kados or fix it :-)
20:48 slef s/committed/& yet/
20:48 chris and then comment it, so i dont go huh?? next time :)
20:48 kados hehe
20:50 slef does SVN do per-branch permissions?
20:50 chris good question, i dont know the answer though
20:51 http://www.fedoraproject.org/w[…]re/VersionControl
20:51 is kinda interesting
20:51 slef I think part of the problem is that the default commit is into the next release, but to do otherwise in CVS (and I guess also SVN) is more work.
20:52 chris right
21:17 slef wow... #savannah reports it was 3-disks of a 6-disk RAID array failed
21:19 rather, 2 disks of a 5-disk+1-hotswap array failed
21:20 chris ouch
21:20 slef that's a bit unlucky
21:22 they're just finishing the hardware config and about to restore from backups
21:23 chris cool
21:23 slef 560GB over gigabit ethernet
21:23 then they have to cart it back to the colo
21:23 I guess they're in Boston MA.  The colo is in Quincy.
21:23 chris ahh, not too far then
21:25 slef My US geography is terrible ;-)
21:26 chris ive been to a wedding in Boston but it was a while ago now
21:27 slef looks like 10mins train if it was here
22:26 savannah 40% copied
22:27 kados so were they able to restore from raid rather than backup?
22:27 slef no, the failure pattern broke the RAID
22:27 kados shit
22:27 so we roll back to Sunday then
22:30 chris kados: you should be able to redo the commits .. from your checkout hopefully
22:30 kados naw, it will think they are already done
22:30 cvs-
22:30 -
22:30 :-)
22:30 chris hmm will it?
22:31 kados yea, the versions are stored in the Entries file
22:31 and they will conflict with what's on savannah
22:31 chris bummer
22:31 slef kados: want an evil way to do it?
22:31 kados hehe, sure
22:32 slef tar it up, with --exclude CVS, get a new checkout on the right branch, untar over it, commit, run away screaming
22:32 kados I'm sure there's some obscene bash onliner that would check out a fresh copy and recursively replace the CVS dir in each of my working copy's dirs with a fresh one
22:32 hehe
22:32 yea, I especially like the part where I get kicked off the project :-)
22:32 slef actually, could be smarter with tar
22:33 kados for completely borking CVS :-)
22:33 slef only tar up files newer than Sunday 2030
22:33 kados I can see it now 'well first savannah was down for two days, then kados broke the repo' :-)
22:33 slef hey, we know they have a backup
22:33 kados hehe
22:34 slef believe me, you want to be the first to commit in the above way
22:35 everyone else will probably have to merge ;-)
22:36 kados yea, that's the prob
22:36 I don't trust that I have the latest version of every dir
22:36 so tell me how this would work with git?
22:37 slef we point and laugh at the failed server, then replace it with the release manager's current local copy (or a mirror of it)
22:37 kados so the working copy comprehends the whole history?
22:38 (I like the point and laugh bit)
22:38 slef rather, the local store does
22:38 so I should have said local store
22:38 kados k
22:38 slef or local repo
22:38 not local copy, which was ambiguous, sorry
22:38 kados np
22:39 slef I'm more accurate before 0000
22:39 kados hehe
22:39 chris :)
22:39 kados hehe
22:39 slef boy, the old test system is slooooow
22:40 and it's the only mipsel now, so I can't distcc
22:40 kados what about the access rules, etc.
22:40 so there's the RM's repo
22:40 people can't commit to that, right?
22:41 slef not unless the RM sets it that way
22:41 kados the RM has to merge code into it from other repos?
22:41 slef it's up to us, really
22:41 kados ok, and renaming dirs, files and changing permissions I assume can be done painlessly
22:42 slef lots of choices... CVS actually also gives some of these choices, but no hosting services support it AFAIK
22:42 chris yeah you can do all the cleaning up that cvs doesnt let you do
22:42 slef git mv, and so on
22:42 kados right
22:43 so if I set up git on one of liblime's servers
22:43 and we pointed git.koha.org to it
22:43 slef not sure about chmod, but the build scripts ought to be doing that IMO
22:43 kados could we use it in the same way we use CVS right now
22:43 where every dev has a ssh key, they have write access to the repo, etc.
22:44 they each check out a version of it, modify, and commit back?
22:44 slef if you set up git-cvsserver, in exactly the same way I think... http://www.kernel.org/pub/soft[…]it-cvsserver.html
22:45 kados CVS clients cannot tag, branch or perform GIT merges.
22:45 slef although AIUI each branch would have its own cvsroot
22:46 CVS clients wouldn't know what a GIT branch is... different level to a CVS branch
22:46 You can CVS branch one file, remember.
22:46 kados yea
22:46 ok, now lets say we set up git
22:46 :-)
22:46 and chris has a repo
22:46 I have a repo
22:47 paul has a repo
22:47 now chris does some work on acquisitions
22:47 and so does paul
22:47 noooo!
22:48 slef typical!
22:48 kados :-)
22:48 anyway
22:48 slef I wouldn't host it here, don't worry!
22:48 kados so paul, chris and I each have our own repos
22:48 slef It would be in London Docklands.
22:48 kados and paul and chris both do work on acquisitions
22:48 in CVS here's what would happen:
22:49 paul commits his code
22:49 chris commits his code
22:49 chris it doesnt let me
22:49 kados he's told to update first
22:49 chris it makes me update then it conflicts
22:49 kados yep
22:49 chris then i have to go fix it
22:49 kados he resolves conflicts
22:49 chris then i commit
22:49 kados and commits
22:49 done
22:49 what's the step by step if we had git
22:49 for getting both paul and chris's code into cvs
22:49 into my repo I mean :-)
22:49 chris and meanwhile paul has committed again and i have to do it again:)
22:49 slef paul commits his code
22:50 chris commits his code
22:50 kados ...
22:50 slef then probably when chris tries to push to paul, it asks him to merge
22:51 kados why is chris pushing to paul?
22:51 slef (but git does better at conflict resolution than cvs IME)
22:51 oh wait
22:51 are both pushing to you?
22:51 kados yea
22:51 slef well, whoever is pushing second may get need to merge
22:51 s/get/
22:52 chris yep
22:52 i could also push to paul if i wanted
22:52 kados hmmm
22:52 slef depends if you have edited the exact same lines incompatibly
22:52 kados hang on
22:52 slef yeah, if kados's server blows up, chris can push to paul
22:52 kados I thought that when you push it becomes a branch
22:53 slef or if chris decides "I don't have time to merge right now" he can quickly mail to paul and try to persuade paul to merge it ;-)
22:53 kados see that's what I'm worried about right there
22:53 chris yeah i can say "paul ive done a bunch of work an acqui, can you git-pull it"
22:53 kados right now, I have made the case to everyone
22:53 somewhat successfuly
22:54 that we all need to commit our work to CVS and stick with the project
22:54 where's the incentive to do all this push-pull with git?
22:54 i dunno if that's a legitimate question or not
22:54 slef well, what's the incentive to do it with CVS?  It doesn't change.
22:54 chris yeah, its a person problem
22:54 kados well with CVS you have to resolve conflicts if you want any kind of version control
22:55 with git you have version control without resolving conflicts
22:55 chris id still have to do that locally
22:55 slef In fact, the wider range of ways to push/pull/mail/clone and so on will probably encourage it.
22:55 chris if i wanted anyone elses code
22:55 slef kados: erm, no, you can run other version control systems on CVS working copies
22:56 chris otherwise its the same as me grabbing the files and running my own local cvs repo
22:56 ie if i didnt want to participate in the project and commit my code back now, cvs isnt forcing me .. i could just do what the south americans did
22:56 kados maybe my idea about how software development should work is the problem here
22:57 chris people have to want to commit back
22:57 kados my core assumption is that it's a good thing that we all have to see each other's code
22:57 slef I want to, but current situation means it doesn't happen soon enough and that makes the task bigger.
22:58 kados esp since there are only about 10 of us
22:58 chris yep, and with git that would still happen
22:58 kados and probably only about 4 that commit regularly
22:58 would we have a koha-git log of every commit?
22:58 chris ie i would pull from your (The release managers) repo
22:59 slef with git it could happen more... I could git-format-patch stuff to email and ask 'Is this worth tidying up into a commit?'
22:59 chris yes git does good logs
23:00 kados I think the biggest problem is that we need to all agree on how exactly our releases should be developed :-)
23:00 i envision the following basic procedure for a stable branch:
23:01 1. build a release
23:01 2. fix some bugs, minor new features
23:01 3. test
23:01 4. build a release
23:01 5. start over at 2
23:01 slef random point: put everything, including roadmaps and so on, into the repo.  There are times (trains, some customer sites) when I can hack but don't have good network access.
23:01 kados in reality, we're a diverse group
23:01 so while I'm at step 2, paul might be at step 3
23:02 slef: yea good point
23:02 chris yep, its all policy
23:02 kados that's where our conflict happens
23:02 slef (but koha is better than many at documenting/commenting... I know at least two projects I use a lot who put 'See http://inaccessible' in the READMEs.)
23:02 kados so right now for instance
23:02 paul is ready for #4
23:02 but I'm still at #2 with 3.0 :-)
23:03 and parts #3
23:03 chris right, thats not a problem git or svn or mecurial could fix
23:03 kados yea
23:03 so maybe the thing we ultimately need to do
23:03 is have better communication between the developers
23:03 slef bring back the town hall?
23:03 kados about a release schedule
23:04 chris yeah, town hall was good too
23:04 kados once 3.0's sufficiently stable
23:04 we shoudl aim for quarterly releases or something
23:04 that follow a consistant pattern
23:04 slef and also I think each RM could take a step back from coding and get more done by getting particular contributions... sort of commissioning
23:04 kados and maybe we can align our schedules so that we're all building at the same time, testing at the same time
23:04 and when it comes time to build the relase, we're all ready for it
23:05 slef: yea, that's nice in theory, but in practice, the RMs are also the lead developers :-)
23:05 slef sad as it may sound, I think we need a developer portal
23:05 chris yeah
23:05 i started that in 2000
23:06 kados it's a receipe for disaster
23:06 I actually didn't intend to be a lead developer :-)
23:06 chris http://developer.koha.org/ i didnt get far
23:06 kados honestly :-)
23:06 chris :)
23:06 actually i lie, that used to have stuff there
23:06 slef anyone want to buy me a new connection to nz
23:06 with a banner on the top "this is build season/ this is new features season/ this is freeze season"
23:06 kados hehe
23:06 chris but it was so old an out of date i took out in about 2003
23:07 good idea
23:07 kados clever
23:07 seasons
23:07 I like it
23:07 slef maybe the wiki could do this most easily
23:07 but I have a love-hate with that wiki
23:07 chris heh
23:07 kados yea, me too
23:07 sometimes it doesn't handle authentication properly
23:07 you too?
23:07 chris yeah i have a love hate with all wiki's
23:07 kados yea
23:08 one season per month
23:08 chris i do think putting the roadmap and things in the repo is a fantastic idea
23:08 kados yea, definitely
23:08 for me, finding a roadmap format is difficult
23:09 it's a tough thing to represent
23:09 chris yep
23:09 kados not like building a house
23:09 where you can actually draw everything
23:09 slef probably more normal to do a simple TODO in the repo
23:09 chris yep
23:09 slef or you could put a dot file in there if you're evil
23:10 chris i think that having a place where ppl update what they are currently working on would help
23:10 chris - acquisitions receiving orders 2007-03-15
23:11 so that when i come along to work, i can look and see, oh hey paul said he is working on that, ill ping him and see what hes up to
23:11 kados right
23:11 chris might be at our portal
23:11 or wiki
23:11 kados dotproject maybe
23:12 chris maybe that
23:12 dewey i heard maybe that was a clue.  I know that data size errors have caused Koha to stop functioning when a value was arbitrarily truncated.
23:12 kados ok, so just a sec
23:12 I just had a random thought about git
23:12 question actually
23:12 so in our scenerio in cvs we have the following commands:
23:12 cvs commit (paul)
23:12 cvs commit (chris)
23:13 cvs update (chris)
23:13 cvs commit (chris)
23:13 what would it be in git?
23:13 git commit (paul)
23:13 git commit (chris)
23:13 git merge (chris)
23:13 git merge (paul)
23:13 resolve conflict
23:13 git commit (paul)
23:13 right?
23:13 slef not quite
23:13 git commit (paul)
23:14 git commit (chris)
23:14 git pull (chris)
23:14 git merge (chris)
23:14 git pull (paul)
23:14 or the equivalent with push
23:14 you missed paul's cvs update off, by the way
23:14 if you mean to leave them with the same latest files
23:15 kados right
23:15 yea, so is it more commands in git than cvs to result in the same state?
23:15 (not that that's the only measure of git's worth :_))
23:15 (just something I thought of)
23:15 chris the conflict resolution/merge is much smarter
23:15 slef should be the same number of operations
23:16 how many commands depends on the nature of the changes and conflicts
23:17 kados I know I could use some better development habits
23:17 like diff and patch are not really my friends
23:18 chris the nice thing with git is it has stuff like
23:18 kados and I always worry when I do a cvs update and cvs merges that while it may have successfully merged code, I have no way of knowing if the result is sematicallly correct
23:18 chris git-send-email(1)
23:18          Send a collection of patches as emails.
23:19 kados ie, I could have a successful cvs update that results in code that doesn't compile
23:19 chris and git-format-patch
23:19 kados or that does wild and unexpected tings :-)
23:19 slef kados: that's more about testing than version control.  You can help that, but not really force it.
23:20 kados yea
23:20 it's a people problem again :-)
23:20 chris: *nod*
23:21 slef one git safeguard that I think is now on by default is that you can name exactly which files get committed, so no more accidental commits... it sounds a pain, but I actually find it useful.
23:21 kados cvs has that too, eh?
23:21 chris the thing too remember with git, merging is as easy as it can be, cos the linux guys get patches out the ying yang
23:21 kados yea
23:21 chris so its all about making their life easier
23:21 slef I have a script that marks any file mentioned in my Changelog for commit... if something I expect to be committed isn't marked, then I've missed it from the Changelog
23:22 kados: if you do cvs commit in a dir, it checks in everything it can find and then some
23:22 not sure if you can stop that
23:22 kados well you can say cvs commit filename1 filename2 etc
23:24 slef: btw, I had no idea that Richard Stallman was so hard to deal with ... based on #savannah sounds like he's pretty stubbrn :-)
23:25 slef erm, he's not hard to deal with
23:25 but he is extremely stubborn
23:25 kados not prioritizing savannah was a big mistake ... a two day outage for hundreds of free software projects could make national news
23:25 slef If RMS has decided, forget changing it, IME.
23:26 kados it's something that microsoft just waits to point to
23:26 slef nah, the news media are still going on about the GPG security problem
23:27 kados hehe
23:27 slef http://www.scmagazine.com/us/n[…]ing-like-attacks/
23:28 kados jeez that's bad
23:28 only affects email?
23:28 slef yep
23:28 kados not too bad then
23:29 but definitely embarassing
23:29 :)
23:29 slef also, GPG users && html mail link clickers && banks using GPG is approximately 0
23:29 kados yea
23:29 slef but don't get me onto banks and secure email
23:30 kados yea, sending statements and such via email is a bad idea
23:30 slef:  do you know anyone at gna?
23:31 I'm curious if they've got redundency
23:32 slef I know people who used to be at gna.
23:35 http://emergency.gna.org/
23:36 https://gna.org/file/diagram.s[…]s.png?file_id=479
23:37 https://gna.org/cookbook/?func[…]litem&item_id=105
23:38 kados slef: what about cvs history, can we import history into git?
23:39 slef http://www.kernel.org/pub/soft[…]vs-migration.html
23:39 in short: yes
23:39 in long: requires additional software, hammers the cvs server
23:40 aside: this is how I run git on top of koha's cvs
23:45 kados slef: on another note
23:46 slef:  what would you say to the idea of doing autoconf/automake for koha 3.0?
23:46 slef eek
23:46 but I see where you're coming from
23:46 set up zebra (and yaz?)
23:46 kados yea, maybe
23:46 just generally clean up the install process
23:47 slef I'm a bit stuck with MakeMaker... it's not really designed for cgi-bin
23:47 kados which is so hard right now
23:48 slef having the directory structure tidied would help MakeMaker
23:48 also I need to find a bit of time to ask debian's pkg-perl people
23:48 kados we're getting there in 3.0
23:48 slef as they must have done this before
23:48 kados directory structure tidying I mean
23:48 slef There are so many options (MakeMaker, Module::Build, PAR, ...) and none seem quite right.  Not sure that autoconf is either tbh.
23:49 kados do you have some time to spend on this?
23:49 tidying up our install process is a major goal of 3.0
23:49 slef Can do next week.
23:49 kados sweet, thx
23:50 slef jabber or mail me if I don't start emailing Mon/Tue
23:50 kados hehe, will do
00:34 slef http://owu.towers.org.uk/planets/koha/
00:34 There's also an index.rss in there if you want to subscribe.
00:34 chris cool :)
00:34 slef call it beta until next week
00:36 I think that could be where we mention "I'm current working on X"
00:36 chris good idea
00:42 sleep well
00:43 slef ta
01:07 chris been trying it out?
01:08 kados not yet
01:08 about to
01:08 we don't really have a server for repos
01:08 chris if you want you can clone my kohabot project
01:08 kados :(
01:08 k
01:09 chris 2 secs lemme set it up
01:10 kados http://code.google.com/soc/
01:11 they still have it down for the 14th :-)
01:11 wonder if people work overtime there
01:12 chris pacific time they have a few hours yet
01:12 kados ahh, right
01:12 hehe
01:16 hmmm
01:16 Git does not explicitly record file revision relationships at any level below the source code tree
01:17 looks like viewing change history on a single file is somewhat complicated
01:18 maybe it's abstracted by the tools though
01:18 chris git log filename
01:18 seems to work fine
01:19 ok wanna try something?
01:20 kados sure
01:24 chris sorry firewall issues, 2 secs
01:29 kados hmmm, there's a git on debian that's different
01:29 it's a file browser/viewser and process viewer/killer :-)
01:29 guess gotta install from source
01:30 chris i didnt, i installed the debian one
01:30 but thats unstable
01:30 kados etch?
01:30 chris debian unstable
01:30 i never know the names
01:31 kados sarge is stable and I think etch is unstable
01:31 is it stable<-unstable<-testing ?
01:31 chris i think etch might be testing
01:31 no stable testing unstable experimental
01:31 kados ahh, right
01:31 yea, so etch is testing
01:35 well I'm too tired :-)
01:35 I better git to bed :-)
01:35 chris :)
01:36 git clone 203.97.214.51:/home/chris/kohabotshared/ kohabot
01:36 if i could beat my firewall into shape
01:36 would get you a copy
01:38 fixed it i think
09:48 kados LibLime was selected for 'Google Summer of Code'
09:48 yay! :-)
09:50 Lea hi guys
09:50 congrats kados
09:50 owen What does that mean kados?
09:50 kados it means that google will pay students to work for us :-)
09:51 on any of the ideas we list on our 'ideas page' :
09:51 http://wiki.liblime.com/doku.p[…]summerofcodeideas
09:51 hdl Hi.
09:51 Wow.
09:51 kados (so if you have some ideas to list, by all means do)
09:51 hdl: is toins still a student?
09:51 hdl: if so, he should sign up
09:51 hdl yes.
09:51 kados good for his resume too
09:51 hdl I shall tell him.
09:51 kados thanks
09:52 hdl He will try to work on opensharetags with three friends of his.
09:52 for a work for University
09:52 kados can someone add the idea to http://wiki.liblime.com/doku.p[…]ummerofcodeideas?
09:52 what is opensharetags?
09:53 hdl A way to share social bookmarks tags wetween ilses.
09:53 Paul's idea.
09:53 kados cool
10:17 owen kados--I was just looking at the Summer of Code page. You're in really good company there!!
10:20 kados yea, it's quite an honor
10:25 hdl: is toins around? we should discuss his idea asap
10:26 PaulShannon: morning .. how'd the tar.gz work out for you?
10:26 PaulShannon Anyone have some time to answer some basic development questions about Koha?
10:33 tnb hdl: do you know anything about Spanish Koha Users Groups?  There's a question on teh mailing list about it, plus I've been wanting to collect information about Spanish-speaking userss
10:34 I seem to remembering stumbling across a site awhile back....
10:55 PaulShannon Does anyone have some sample bib data I can dump in for evaluation?

← Previous day | Today | Next day → | Search | Index

koha1