Camelia, the Perl 6 bug

IRC log for #darcs, 2010-06-08

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:32 lispy joined #darcs
01:53 npouillard joined #darcs
02:03 zooko joined #darcs
02:39 kpreid_ joined #darcs
03:44 abuiles joined #darcs
04:05 twb joined #darcs
05:18 abuiles1 joined #darcs
05:43 balor joined #darcs
05:52 mornfall *sigh* OSX
05:54 mornfall You can't win, it's either win32 or osx that has to break.
05:55 mornfall Can anyone with OSX try if date -f "now-350days" works on there? Or something equivalent...
05:56 mornfall Implementing date arithmetic in bash would be stupid.
06:02 twb date isn't bash
06:03 lispy $ date -f "now-350days"
06:03 lispy usage: date [-jnu] [-d dst] [-r seconds] [-t west] [-v[+|-]val[ymwdHMS]] ...
06:03 lispy [-f fmt date | [[[mm]dd]HH]MM[[cc]yy][.ss]] [+format]
06:03 twb Does OS X have /usr/gnu/bin/date or similar? ;-)
06:03 twb Stupid shitty BSD userlands...
06:04 mornfall Well, all I need is a date "bit less than year ago".
06:04 mornfall In the testsuite.
06:05 mornfall The other option is probably doing it in Haskell....
06:05 twb you probably don't have @<epoch time> either
06:06 lispy $ date -v -350d
06:06 lispy Mon Jun 22 23:06:34 PDT 2009
06:06 mornfall \o/
06:07 mornfall Is that documented at all? (D'oh...)
06:07 lispy yeah, man date
06:07 mornfall Ok, thanks.
06:07 lispy "There's a mandate?"
06:08 * lispy >>= bed
06:08 mornfall Night.
06:08 twb That format isn't valid in GNU date
06:09 mornfall twb: Obviously, but I can do some if-ing and kicking to get a date out of either.
06:09 twb OK
06:09 mornfall lispy: (Does it also work with +fmt?)
06:09 lispy you mean, +350d?
06:09 lispy if so, yes
06:09 mornfall With date -v -350d +%Y%m%d
06:10 lispy oh, yes
06:10 mornfall Ok, thanks.
06:10 lispy 20090622
06:10 mornfall Zuper.
06:21 lelit joined #darcs
07:02 darcscommitbot joined #darcs
07:03 darcswikibot joined #darcs
07:41 skade joined #darcs
07:46 sxs joined #darcs
08:04 mornfall_ joined #darcs
08:05 aavogt1 joined #darcs
08:08 twb joined #darcs
08:32 mornfall joined #darcs
08:35 balor joined #darcs
08:40 sxs hi
08:40 sxs why is it possable to push a patch more than one time into a repo?
08:41 twb Define "more than one time"
08:41 twb push should be idempotent for any given patch.
08:41 twb (Note that amend and rollback, at least, actually create entirely new patches.)
08:41 sxs and wenn the patch differs?
08:42 sxs hm, yes i think its a problem of the amending
08:43 sxs we have an integration repo. and we have the rule, that any developer must push to tghis repo bevor we deploy. sometimes a developer changes his patch and forget to unpull his patch befor pushing again.
08:43 sxs then we have the patch twice in the repo
08:45 kowey joined #darcs
08:45 kowey morning
08:45 twb sxs: the solution is basically: don't do that, then
08:47 kowey a good rule to implement would be once you've pushed to that integration repo (just skimmed the chat log), no more amending or unpulling
08:47 twb Yep
08:48 twb To help me, what I usually do is record patches with the name "DRAFT: foo" until they're ready to push
08:48 kowey you can also check to see if it's safe to amend a particular patch by doing darcs push --dry-run
08:48 twb And tell amend-record to only match 'name DRAFT and author Trent'
08:48 twb kowey: yeah, but that's a pain in the arse
08:48 kowey the DRAFT technique is pretty handy too
08:49 twb It's a bit of a bodge, but it Works For MeTM
08:50 gal_bolle joined #darcs
08:50 kowey it's conceivable that there be a feature request for darcs amend --against-repo X
08:51 twb More generally, it's an argument for Darcs knowing about more than just a single branch of a single repo
08:51 kowey which would either not let you amend patches already in X, or offer to record a new one
08:51 twb It's unlikely that I'd ever use an --against-repo argument
08:52 twb If patch metadata had some kind of "published" bit, I might use push --set-publish
08:52 gal_bolle i'm not sure what you're talking about, but I think a "in [repo]" matcher would be very useful
08:52 kowey http://bugs.darcs.net/issue1613 for reference (feature request for a notion of patch annotations)
08:52 gal_bolle so that you can say darcs amend --match "not in upstream" or darcs unpull --match "in upstream"
08:53 kowey gal_bolle: I'm trying to think of the scenario "is it safe for me to amend this patch?"
08:53 sxs we are using special IDs for our Patches like 2347_1 which means first patch fir jobticket 2347
08:53 gal_bolle kowey that would be the use case for my 'in' matcher, then
08:53 kowey gal_bolle: you can check manually by doing darcs push --dry-run (only amend *those* patches)
08:53 gal_bolle k
08:53 sxs is there something like a hash which identifies a patch unique? so that i can compare twho patches with same name if they differ?
08:54 gal_bolle kowey: darcs is supposed to let you express the right thing, not manually do things
08:54 kowey gal_bolle: but I was trying to see if there was a way Darcs could be safer/more automatic
08:54 kowey jinx :-)
08:54 gal_bolle i think --match "in" is the darcsishly smart
08:55 gal_bolle (not addition of external knowledge, but lets you express the safety you want, orthogonal and symmetric)
08:55 kowey it'd involve some changes to the match infrastructure, presumably, but that's no reason not to do it...
08:56 kowey sxs: yes, see http://wiki.darcs.net/Hashes
08:56 gal_bolle that's something i'd volunteer to implement
08:57 kowey sxs: if you amend a patch, the patch info hash will change because the date changes (and maybe the patch salt, but I forget)
09:04 sxs another problem is, why we do a lot of amending is that darcs slows down on a large amount of patches :/
09:04 gal_bolle in the interactive selection?
09:05 gal_bolle with what command/options?
09:05 sxs i never tested, I'm new in the company. But my boss told that it was neccessary to join some old patches (few thousand patches) to one big patch
09:06 sxs i know this phenomenon from SVN. above a view thousand commits the whole system slows down
09:06 gal_bolle i'm not sure I understand what you want to do
09:07 gal_bolle you want to unrecord a bunch of old patches and re-record them together?
09:09 twb I think Darcs is likely to have more problems with a thousands-of-lines commit than a thousand small commits.
09:14 sxs I did not remember how my boss "merges" the patches. but what iI can see everyday is that darcs is not tthat fast on large repos witch > 10.000 patches
09:14 sxs so we are adviced from our boss to make large patches than a lot of small ones
09:14 kowey it's a good idea to tag fairly regularly (for example, the darcs team do so on every release)
09:14 gal_bolle what command[s] do you experience slowness with?
09:15 kowey (but yes, this is very useful performance report, thanks!)
09:15 twb And to use darcs-2 format instead of old-fashioned
09:16 sxs i usualy use pull/push/get (i'm something like  release manager) i use record seldom
09:16 sxs we are using darcs-2 format
09:16 sxs but an older version of darcs
09:16 sxs 2.2.0
09:17 gal_bolle do you pull/cherry-pick from repositories with lots of new patches?
09:17 twb I doubt performance has increased drastically between 2.2 and 2.4
09:19 sxs gal_bolle: not really. What I do: I have a jobticket (eg. 42). Then i pick all patches from the developer which did the ticket (42_1, 42_2 etc.) amd push it to alpha, then testing, then push to beta, testing a.s.o.
09:20 twb Here, each ticket would normally only get one or two patches
09:20 sxs the biggest problem IMHO is the lack of practice with distributed VCS :(
09:20 gal_bolle do you pull them from a repository that has only these patches, or one that has all the programmer's work?
09:20 sxs gal_bolle: from one that has all the programmers work
09:20 gal_bolle ok
09:21 twb But we also test each ticket being fixed in its own branch, before pushing to a central area
09:21 twb So usually if it's obviously wrong (e.g. fails to compile), it would be amended, rather than making a new patch
09:22 gal_bolle is it the interactive pull or the application that is slow? (do you wait a long time before you are asked about the patches, or after, or both?)
09:22 gal_bolle (or for each patch you get asked about)
09:24 mornfall twb: It has improved drastically between 2.4 and 2.5-to-be-alpha, though. :)
09:24 sxs it takes time before it askes my first time. its like dars thinks a lot about something and then starts asking for each patch
09:24 gal_bolle ok, then it could get better in 2.5
09:25 sxs twb: yes of course. alpha/beta is for in the wild test. We are dirty PHP-Hackers ;) alpha/beta means that our real customers test the usability
09:25 twb That's "integration testing"
09:25 gal_bolle sxs: do you use --match?
09:25 twb And I'd say that if you're using PHP, you're probably screwed no matter what you do with Darcs.
09:26 sxs twb: Really? I thought it was acceptance test, when beta users test the usability etc.
09:26 mornfall :)
09:26 sxs gal_bolle: no, i'm not so familiar with darcs. what can --match do for me?
09:26 twb integration tests happen after unit tests and before acceptance tests
09:27 gal_bolle you can use it to pre-sort the patches you're going to pull
09:27 mornfall and after functional tests, presumably
09:27 mornfall I'd say pre-filter.
09:27 mornfall (About match...)
09:27 twb Depends on your definition of function.  If you're testing each procedure individually, that would go with/before unit testing.
09:27 sxs twb: Yes, we are screwed. but you can't change the language after the project runs for 10 years :/ i dont like PHP.
09:27 gal_bolle for instance, you can say darcs pull --match "author joe" to pull joe's patches
09:28 gal_bolle it's prefiltering
09:28 twb sxs: you could always go flip burgers.  You might be poor, but at least you wouldn't hate yourself
09:28 mornfall twb: wp functional testing (they seem to call it "system testing" though...)
09:28 mornfall It also seems that I misunderstand what "integration testing" means. :D
09:29 sxs mornfall: these are all buzzwords :P
09:29 sxs match an author would be an interesting thin.
09:29 twb mornfall: integration means "did all the bits fit together?"
09:29 mornfall twb: Yes, but I sort of expected integration of the product with outer world.
09:29 twb mornfall: like, suppose you test the jet engine and the wing separately (unit tests), and then you connect them together and test both at once (integration).
09:30 mornfall It's all fuzzy.
09:30 mornfall I distinguish unit testing (Darcs.Patch.Core) and functional testing (darcs executable).
09:31 mornfall tests/* is functional, QC and HUnit are unit
09:32 mornfall It seems that integration testing is just unit testing of bigger units. It would, in this analogy, test that Darcs.Patch works with Darcs.Repository or somesuch.
09:32 sxs IMHO unittesting tests a unit (class, method, function package) whatever you define as unit. functional testing: you test the whole functionality (in scrum speak the story). Integration tests are tests like unittests without mocking, through the whole system. Acceptance tests are enduser tests.
09:37 gal_bolle sxs: with current HEAD, in an empty repository, pulling from the ghc darcs repository starts asking me questions right away (there's a ~ 2s network delay, then it starts). Then selecting a few patches and applying them is reasonably fast
09:43 sxs hm. ok. the "performance" is not our real problem. as mentioned above: our biggest problem is that the most developers are not familiar with distributed vcs. the most arent familiar with an stupid tool like svn :/
09:44 sxs and i try to make some tools/rules to prohibit fuckups
09:44 sxs how can i display the patchs hash?
09:44 mornfall darcs chan --xml
09:44 skade left #darcs
09:45 sxs ahhhh
09:45 sxs thx
09:46 sxs is this from darcs? <comment>Ignore-this: e27fd9b3e49a02dd18946ecdccb2852f</comment>
09:46 sxs we do not use the long comment
09:48 twb sxs: yes.  It's there specifically to annoy me.
09:49 sxs so darcs insert this ignor-this-foo, unless i input a long comment
09:51 gal_bolle it does it whether or not there is a long comment. But darcs 2.3+ (i thin does not display it
09:52 twb sxs: it always inserts it
10:13 kowey resolve issue8917: annoy Trent
10:14 kowey it's actually there for safety reasons
10:14 kowey I'm not happy about it being generated from random noise, but we don't yet have the time to make it non-random
10:32 twb It's because normal people value functionality over beauty.
10:36 Korusef joined #darcs
10:54 gh_ joined #darcs
10:57 intripoon joined #darcs
11:01 mornfall Some tests failed:
11:01 mornfall issue1210-no-global-cache-in-sources.sh
11:01 mornfall (win32)
11:01 mornfall lambdabot: @tell abuiles issue1210 test fails on Windows, see http://buildbot.darcs.net/builders/6.10.3​%20Vista/builds/25/steps/test/logs/stdio
11:01 lambdabot Consider it noted.
12:08 markstos joined #darcs
12:10 skade joined #darcs
12:44 bonobo joined #darcs
12:49 skade joined #darcs
13:12 zooko joined #darcs
13:32 skade joined #darcs
13:51 felipe joined #darcs
13:53 skade joined #darcs
14:00 skade joined #darcs
14:47 abuiles joined #darcs
14:48 abuiles mornfall, I was afraid of that one..
14:48 lambdabot abuiles: You have 1 new message. '/msg lambdabot @messages' to read it.
15:06 abuiles left #darcs
15:51 zooko joined #darcs
16:22 kowey mornfall: hmm, this looks interesting re windows and filepaths, http://www.haskell.org/pipermail/h​askell-cafe/2009-June/062919.html (but sounds like what you already know)
16:32 kowey on that thread, Judah also pointed to System.Console.Haskeline.Directory
16:33 kowey ...wonder if it could make our lives a bit easier... http://code.haskell.org/haskeline/Sy​stem/Console/Haskeline/Directory.hsc
17:35 www joined #darcs
17:37 abuiles joined #darcs
18:51 abuiles mornfall, are the test in windows run under cygwin ?
18:58 tux_rocker joined #darcs
19:01 arjanb joined #darcs
19:07 kowey joined #darcs
19:09 _ilbot2 joined #darcs
19:09 Topic for #darcsis now can't talk? see -> | http://wiki.darcs.net/IRC | http://darcs.net/ | latest is 2.4.4
19:11 tux_rocker there's 278 unread emails in my "darcs" folder
19:11 tux_rocker hmmm
19:11 tux_rocker how do you approach all the mail coming down the darcs-devel and darcs-users pipes?
19:12 tux_rocker how do you filter it? what do you read, and what do you leave for others to handle?
19:12 Heffalump I just don't read it :-)
19:27 mornfall abuiles: No.
19:28 kowey tux_rocker: you're the 3rd Darcs hacker to say that!
19:29 kowey hmm
19:30 kowey so darcs-devel is pretty much something that can be automatically filed away
19:30 kowey it's kind of useful to see the subject line pass by and have it seep into your unconscious, imho
19:32 abuiles mornfall, I amend the patch, I'm skipping the test in windows.
19:32 abuiles hi kowey
19:32 kowey is it clear yet why it fails under Windows?
19:33 abuiles I think is because the location of the global cache. in windows is not $HOME.
19:33 kowey so why does the the ${HOME} entry appear in the cache at all?
19:33 mornfall abuiles: That shouldn't matter.
19:34 mornfall cache:/c/buildbot/darcs/6.10.3 Vista/build/tests-darcs-2.dir/.darcs/cache
19:34 mornfall This should not be in the sources.
19:34 mornfall And it is, on win32... why?
19:35 abuiles so, it wasn't what I was thinking...
19:36 abuiles that cache shouldn't  be there...
19:36 kowey tux_rocker: if I may make the suggestion, two key things to watch out for are (a) things assigned to you [should be to:you specifically] and (b) DWN - so you have a general sense of what's going on
19:36 tux_rocker kowey: i thought of those already, yeah
19:37 kowey threaded mail may help too
19:37 kowey the information overload problem worries me :-/
19:37 tux_rocker kowey: threaded mail in gmail helps
19:37 tux_rocker but then i'm not sure if i like web-based email
19:38 kowey I use gmail via mutt, personally
19:38 kowey but I'm grumpy that way :-)
19:38 tux_rocker KMail, which I use on the desktop, has threaded mail but screws it up
19:38 kowey sup provides gmail-style conversations with a nice text-based interface
19:38 mornfall Hm. I just finished Gobliiins. :)
19:38 tux_rocker afaik you can't sort by most recent post in thread, so you may have new email all the way down there in the list
19:39 mornfall kowey: But sup is no longer a MUA.
19:39 kowey that's one of the things my mutt config does, fwiw -- I suspect that sup would be configurable in that sense too
19:39 mornfall But, I am on gnus which is primarily NNTP and it does good job threading. But you first need to be Emacs user for that to make any sense.
19:39 kowey mornfall: mmm? you mean in the do only one thing well sense or something else?
19:39 mornfall kowey: I mean that the author decided he likes to work on a document storage system more than on a MUA.
19:40 balor joined #darcs
19:40 mornfall kowey: So I think they drop'd IMAP and similar stuff and are working on non-MUA features.
19:41 kowey hmm, interesting
19:41 kowey I'm not sure what MUA features I care about (or what non-MUA features I don't care about) enough to see the implications of this
19:41 kowey tux_rocker: another trick I use, although you already get this for free with gmail, is just to have one folder called 'archive'
19:42 kowey without trying to be clever about organising mail
19:42 kowey this reduces the friction involved in getting things out of that inbox (I could go on all day, so  should probably stop before people start throwing things at me)
19:43 kowey there's this http://producingoss.com/en/common-pitfalls.html which I may need to be more worried about
19:44 mornfall There’s another convenient benefit to this transformation: no one will expect STS to act like a MUA. STS does its own storage. You add your email and your other documents to the server and then you can throw those files away (or not). There are no more questions of supporting IMAP or various mbox dialects or “why doesn’t Sup treat Maildir correctly”. The files are in STS, and once they’re their, they’re out of your hands.
19:44 mornfall kowey: ^^ This, here.
19:45 mornfall So sup is out of question for me, as long as I do not want to lock myself in. IMAP it is.
19:45 kowey oh :-/
19:45 kowey I think I have many other reasons to use Maildir (for example), eg. Unison
19:45 kowey oh well! no sup for me
19:46 mornfall Hm. About RM-ing.
19:47 kowey [actually, the 'one-archive-folder' thing is a lie; I have procmail squirrel away everything that goes to a mailing list (except darcs) to lists/foo]
19:47 kowey yeah, about RM-ing
19:47 mornfall I think a good RM will actively fix stuff that's broken for release. Win32 and OSX included.
19:47 mornfall And that needs a lot of knowledge.
19:47 mornfall The experience is that if RM does not step up, no-one does.
19:48 kowey so we need a darcs-hacking RM?
19:48 mornfall Well...
19:48 mornfall The catch is, that darcs as a team is very liberal.
19:49 mornfall And you can't command people around.
19:49 mornfall And release-fixing is boring.
19:49 mornfall So being RM mostly means volunteering to do that boring stuff.
19:50 mornfall Sense of responsibility for a release is probably one of the few things that can move hackers to do that work.
19:50 mornfall (And any amount of prodding from RM is unlikely to...)
19:50 kowey well, the RM prodding is mostly an admin thing
19:50 mornfall At least as long as your RM sits a few countries away from you.
19:51 kowey in the sense of preventing deadlock or things falling through the cracks, etc
19:51 kowey you need somebody with sort of a global picture of the release state
19:53 kowey OK, so a good RM should be willing to fix near-release bugs?
19:53 mornfall Well, you know it's not an easy job. There's a reason no-one on the review team wants it.
19:54 kowey well, so if RM is the kind of job that nobody wants, then I should definitely take it
19:54 kowey but this means that something else has to give
19:54 kowey (or maybe not, maybe I'm being too pessimistic)
19:54 mornfall I think managing issues is the same bag.
19:54 mornfall Just different work.
19:54 kowey just a different kind of nobody wants to do it?
19:54 mornfall Yeah. :)
19:55 mornfall Unless you are a gardener, managing a bugtracker is the definition of boring.
19:55 mornfall RM-ing is probably less boring but more stressful and quite unrewarding.
19:55 kowey I can more or less keep on top of that, although what tends to happen is that something will distract me and I'll forget
19:56 kowey the IM stuff would be fine (IMHO) if I could just make a daily habit of it
19:56 mornfall Well, one option is to diffuse the RM duties to the review team as a whole.
19:56 kowey OK, but if we do that, how do we make sure things still get done?
19:57 mornfall Well, the same as if we had a dummy RM. :)
19:57 kowey so responsibilities include: pulling to stable (who makes the call), making freeze announcements, writing ChangeLog entries
19:57 abuiles well, I dunno what's going on with that in windows, I will build it there, and see why is the globalcache being saved.
19:57 kowey in a sense, we've already automated a good chunk of the work - I think we're at a point where we can copy+paste much of the freeze announcements, for example
19:58 kowey abuiles: yeah, if you have a windows machine on hand, that'd be a good step
19:58 kowey abuiles: this is actually good OS experience in the sense of getting a good intuitive grasp of why projects seem to always overrun :-)
19:58 mornfall kowey: Well, we could adopt a slightly different process...
19:58 kowey OS here meaning open source
19:58 * kowey listens
19:59 abuiles kowey: I have a virtual, but building darcs is giving me a headache..
19:59 kowey it's not just install haskell platform and cabal install -f-curl?
19:59 mornfall kowey: Say, stable needs 2 review team ACKs (mainline needs just one).
20:00 kowey heh, we could use David's pull --intersection idea
20:00 mornfall kowey: Reviewers also get a veto on stable inclusion, which if raised will seek consensus.
20:00 mornfall kowey: that's technicality
20:01 mornfall We can do that if communication turns out to be problematic.
20:01 mornfall Anyway, the other thing is freeze plan.
20:01 kowey is that pretty much just automatic anyway?
20:01 mornfall I guess we can just use the template I started a while back.
20:01 kowey we want to release by X, so we just go for X-7, etc?
20:01 mornfall Yeah, something like that.
20:02 * tux_rocker sees an interesting discussion but goes to bed anyway because he even has to vote before going to work tomorrow
20:02 kowey http://wiki.darcs.net/Deve​lopment/ReleaseManagement is current discussion
20:02 kowey good night!
20:02 mornfall Night.
20:02 kowey s/discussion/version/
20:03 abuiles kowey: yeap the -f-curl saved me for going nuts :0, and stop me for hating more windows.. :)
20:04 mornfall kowey: Well, even a bit differently.
20:04 mornfall kowey: We should set soft freeze as the date of the last alpha release.
20:04 kowey now that we're planning on alphas
20:05 mornfall I think that alpha happens when the respective tag gets pushed onto mainline (usual rules applying -- one RT member needs to send it, another needs to push it...).
20:06 mornfall Someone can then make and upload tarballs to Hackage or something.
20:06 mornfall Is there a reason to have non-hackage tars somewhere?
20:06 kowey ...just a quick aside... shall I still try to recruit an RM while we think about this?
20:07 mornfall I don't know. What do you think?
20:07 kowey I don't really think so; we could just make darcs.net point to hackage, but it's the same tarball anyway and seems to cost nothing
20:07 mornfall I think that anyone outside of the review team is not qualified enough for a darcs RM.
20:09 kowey so there are some things like uploading tarballs and writing ChangeLogs that don't actually require that much experience
20:09 mornfall Well, you still need to know what you are talking about when writing changelog summary.
20:09 kowey and also the admin side of things (keeping on top of things)
20:10 kowey that's true (one thing the RM has to be careful to do is to distinguish regressions since last release from actual bug fixes)
20:10 kowey 140 bugs fixed!  :-)
20:10 mornfall admin side can go with you as the issue manager -- just keeping the Target-X.Y bits around should do
20:10 mornfall We should have a policy that release X.Y happens when there have been no open Target-X.Y bugs for a week.
20:11 kowey it's interesting actually
20:11 kowey policy is a form of automation
20:11 mornfall So I guess the Issue Manager should hold the semaphore lights for a given release.
20:11 tux_rocker mornfall: are you inventing this here or has this worked for some projects?
20:12 tux_rocker kowey: i'd say policy is a form of laziness
20:12 tux_rocker and btw, i'm still assuming i'll have to release-manage the 2.5 release
20:12 mornfall tux_rocker: I am improvising a bit. I do take inspiration from working projects, though.
20:12 kowey maybe this needs to become a wider discussion (with some steering) on the list
20:12 kowey not to add to everybody's information overload though
20:13 mornfall I guess we can have an informal vote among reviewers (not trying to form cabals).
20:13 kowey well, we want to be very very careful that things happen in public
20:13 kowey because it's very easy to give people the impression of forming a cabal even if unintentional
20:13 mornfall Well, *this* is public.
20:14 kowey right
20:14 mornfall There's a line somewhere in involving the wider community.
20:14 kowey but this is fundamental project governance, right? seems like a pretty big deal
20:15 kowey (steering would mean making sure we have a quite well specced plan first so that the discussion doesn't derail)
20:15 mornfall I would say that an announcement would be adequate.
20:15 mornfall All the former RMs are on review team, I think?
20:16 kowey unless you want to count Tommy P
20:16 kowey who was the darcs-stable maintainer back in the day
20:16 mornfall Right. And David.
20:16 kowey yeah
20:16 kowey well, everybody who was an RM served on the equivalent of the review team at the time
20:18 kowey OK, so it looks like the worst case scenario is that we don't figure this out in time for 2.5, which means Reinier does the job
20:18 mornfall Actually I don't even think that we need a discussion, nor an announcement.
20:18 mornfall We can just make things happen and discuss along the way.
20:18 mornfall Whatever needs discussing.
20:18 kowey and so far the alternatives are to (A) try and recruit somebody or (B) switch to a more split up model
20:20 mornfall Afterall, if we appointed an RM, who would be making that decision?
20:20 kowey I really hate noise, but I think it's important that things be discussed
20:20 kowey ultimately it's still a dictatorship (answering your question)
20:21 kowey but it's critical that things do not appear to just happen by fiat
20:22 kowey so it's not that the wider community gets a vote
20:22 kowey it's more that they should feel like they can influence the thought process
20:23 mornfall I don't know.
20:23 mornfall Have you noticed, about the alpha?
20:24 mornfall I like when things happen like that.
20:24 kowey what do you mean?
20:24 mornfall It was never discussed on darcs-users@.
20:25 kowey that's true; it was discussed in Zurich among devs
20:25 kowey same with things like the default switches
20:26 mornfall If the discussion wants to happen, it will, even if we don't start it.
20:26 mornfall If people are unhappy about review team governance, they will tell us about it, presumably.
20:27 kowey yeah, I definitely appreciate your low-friction, anti-BS, anti bureaucracy, high-efficiency point of view
20:28 kowey but I have a sort of fuzzier way of looking at things which I have a very hard time explaining clearly (sorry!)
20:28 kowey where it's important in the bigger-picture, longer-term that certain things happen, even if they seem inefficient in the short term
20:28 mornfall You would be a sucky dictator, you mean? :)
20:29 kowey I don't think most dictators feel the need to apologise when they get out the Iron Fist...
20:29 * kowey is too much of a hippie
20:31 kowey OK, I need to go be unproductive for a while... hopefully tonight all this stuff will churn in our heads and  turn into sensible action tomorrow
20:31 kowey good night!
20:31 mornfall I leave tomorrow morning.
20:31 mornfall Be back on weekend or so.
20:31 mornfall Try to shuffle it in your head, turn down the hippie-ness and such.
20:32 kowey ok, I *may* end up starting a discussion (and being careful to make sure I get your position across clearly)
20:32 kowey the good news is you can miss it!
20:32 mornfall It's great to have political speeches in front of darcs-users@, but don't overdo it.
20:32 mornfall Oh, I certainly will. And I'll do what I please no matter its result.
20:33 kowey (oh, I meant as in the timing, like you can come back on the weekend and just select the whole thread in one swoop)
20:33 gh_ joined #darcs
20:33 kowey but, err yeah... we'll work something out
20:34 mornfall Yes, I understood.
20:34 kowey k
20:40 galbolle joined #darcs
20:42 Heffalump gal_bolle: rebase won't be in 2.5
20:43 kpreid_ joined #darcs
20:49 dino- joined #darcs
20:53 abuiles mornfall, I just run the test in my windows machine and passed .
21:17 Iago joined #darcs
21:30 zooko joined #darcs
21:30 zooko I wish for a darcs plugin for
21:30 zooko Hudson.
21:58 abuiles left #darcs
22:36 gwern joined #darcs
23:47 zooko joined #darcs

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary