Camelia, the Perl 6 bug

IRC log for #darcs, 2013-09-09

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

All times shown according to UTC.

Time Nick Message
00:18 sm I contacted some spammy-looking darcs hub users asking them to confirm humanity
00:19 sm now after a suitable period I'm taking the next step, which is to archive their accounts for a while (in case they show up)
00:20 sm I've removed them from user lists and disabled login by giving them an extra "archived" flag, and checking for that in all user db views, in couchdb
00:21 sm but, their public repos still show up in the repo list (though browsing them no longer works)
00:22 sm I wonder what's the simplest way to hide all a user's repos, and what other content be handled the same way. Eg, issues they created ?
00:23 * sm would like to check for the owner's archived flag in the repo db views, but is not sure how to do this kind of "join" query in couch
00:35 favonia joined #darcs
00:43 sm ha.. now I totally break production by messing with couch
00:43 sm serves me right
00:50 arpunk joined #darcs
00:59 sm fixed, with 16 minutes downtime >:|
01:21 mizu_no_oto joined #darcs
01:56 intripoon_ joined #darcs
02:08 carter sm i'm human right?
02:09 carter which darcs hub is the main one?
02:11 mizu_no_oto joined #darcs
02:58 mizu_no_oto joined #darcs
03:25 preflex joined #darcs
04:10 konundra joined #darcs
04:12 favonia joined #darcs
04:16 bsrk joined #darcs
05:17 Heffalump bsrk: sorry I'm late. How's it going?
05:18 bsrk_ joined #darcs
05:18 bsrk_ Hi
05:19 arpunk joined #darcs
05:19 bsrk_ No problem, we also finished the puja just a little late.
05:20 Heffalump I'm just trying your test patch now
05:20 bsrk_ Today is Ganesh Chaturthi. :-)
05:20 bsrk_ Okay, how is it working?
05:20 Heffalump dunno yet :-)
05:21 Heffalump I like tha tthe --oauth tests are off by default
05:24 bsrk_ I expect the changes test will fail - I just noticed and error
05:24 Heffalump ok
05:24 bsrk_ but the other three should pass
05:24 Heffalump you mentioned dropping the authenticate bundles thing - I think that's ok if sm agrees, but is there a way to submit bundles without having an account yet?
05:25 bsrk_ No, there is not one.
05:25 Heffalump yes, 3 passes and failure in the changes test
05:26 bsrk_ is it because it was looking for a link "changes"
05:26 bsrk_ ?
05:26 Heffalump Submitting patches without an account is pretty important, IMO: otherwise random users still don't have an easy way to send a patch
05:26 bsrk_ All right
05:26 bsrk_ What is your opinion on the issue tracker?
05:26 bsrk_ Should it be just there, and there is no manual on/off?
05:26 Heffalump it says Unable to locate element: {"method":"css selector","selector":"#r2 > option[value=\"pnhkehze\"]"}
05:27 Heffalump I just sent an email - I think it has to be turnable off
05:27 bsrk_ oh
05:28 bsrk_ I see
05:28 Heffalump let's see what sm thinks
05:28 Heffalump forwarding to the owner is probably reasonable
05:29 bsrk_ is it? it does not seem helpful in the case of ghc
05:29 Heffalump oh, good point
05:30 Heffalump ideally you want a message from the attempt saying "please send here instead"
05:30 Heffalump but sm is right that special cases etc are something to avoid if possible
05:31 bsrk_ We could keep an error box for send, where if the issue tracker is not enabled,
05:31 bsrk_ it shows the text in the error box
05:32 bsrk_ and perhaps issue tracker should be enabled from now on as defualt
05:33 Heffalump that sounds reasonable to me
05:33 Heffalump can you reply to the thread with that suggestion?
05:33 bsrk_ okay
05:35 Heffalump thanks for updating the todo list, btw - for the last couple of weeks, I think it might also be useful if you could number the outstanding items in priority order
05:35 bsrk_ sure
05:35 Heffalump ok, anything else to discuss?
05:36 bsrk_ I'll add send for random users to the todo
05:36 Heffalump make sure you talk to sm about the authentication thing as well to check he's ok with the behaviour it has now
05:37 Heffalump unless you laready did, I forget :-)
05:37 bsrk_ okay, I'll add this to the discussions
05:38 Heffalump when would you like to meet next?
05:38 bsrk_ How about Thursday?
05:38 Heffalump ok
06:03 intripoon joined #darcs
06:04 dolio joined #darcs
07:06 raichoo joined #darcs
07:44 amgarchIn9 joined #darcs
08:19 schlaftier joined #darcs
08:52 raichoo joined #darcs
09:21 owst joined #darcs
09:21 karlpinc` joined #darcs
10:18 arpunk joined #darcs
10:33 donri joined #darcs
11:04 preflex joined #darcs
13:07 bfrank joined #darcs
13:54 favonia joined #darcs
14:29 sm g'day all
14:30 sm carter: yes! you are human :)
14:30 sm the main darcs hub is supposed to be obvious, please tell me if/why it's not
14:33 sm aha, happy Ganesh Chaturthi all
14:33 favonia joined #darcs
14:33 favonia joined #darcs
14:42 sm so, I have some confusion about workflows and rebase I'd like to discuss
14:43 sm I watched "The Flow of Change" tech talk on youtube, it is a nice basic guide to branch management
14:44 sm and now I think of rebasing (conforming your branch to a new baseline branch) as a normal everyday operation
14:45 sm but as a darcs user it's something I've avoided except in extraordinary cases
14:47 sm this comes up now because: I've avoided working on darcsden several times for fear of the hassle it cause Aditya (forcing him to rebase all his pending patches in Aditya/darcsden-gsoc)
14:47 sm would cause
14:48 sm at least in my current darcs-fu, it would be a big hassle for him. But maybe the rebase command makes it easy now ?
14:48 sm it seems to me it should be(come) as easy as darcs pull --rebase
14:49 sm like git. And, could pull offer to do that automatically when needed ?
14:49 sm And, can that every succeed without manual intervention, or does it always involve some human conflict resolution ?
14:50 sm And, how could this be offered on darcs hub ? So Aditya could click "rebase from upstream" as easily as I can "merge from downstream"
14:51 sm s/every/ever/
14:55 sm secondly, what is the best way to clean up history, eg on a dev branch before pushing to trunk ? Does the rebase command let us easily/automatically merge two named patches into one, for example ?
14:55 sm and can that be done while preserving original patch authorship ?
14:57 sm and, it seems we need more rebase documentation, eg in the manual - I think there is only rebase --help
14:58 * sm waits for Heffalump
15:15 schlaftier joined #darcs
15:23 favonia joined #darcs
15:45 mizu_no_oto joined #darcs
16:07 amgarchIn9 joined #darcs
16:09 mizu_no_oto joined #darcs
16:34 whaletechno joined #darcs
16:39 * Heffalump appears
16:39 int-e yikes
16:39 Heffalump sm: I think you're excessively afraid of conflicts, TBH
16:39 Heffalump darcs is certainly not great at handling them
16:40 Heffalump but small onse are generally ok and managemable
16:40 Heffalump for me, rebase is for substantial issues
16:40 Heffalump rebase is a fallback position. Everytime he did it, I had to mess around with my local checkouts of his work.
16:46 sm Heffalump: if you have 10 or 20 patches pending, and upstream makes a global change conflicting conflicting with all of them, how do you rebase those patches to they'll apply cleanly ?
16:46 sm so
16:46 mtp "freebased VCS" sounds like a designer drug
16:46 mtp ;)
16:58 raichoo joined #darcs
17:28 Heffalump sm: well, the first question for me is *whether* to rebase, not *how* to rebase
17:33 sm Heffalump: well, say I want to avoid conflcts in my repo. Then I need to ask Aditya (say) who has long-pending patches to rebase them against latest trunk before I merge
17:34 sm if I had a large project with many long-lived contributor branches, this is certainly the policy I'd want
17:35 sm even in this small two-branch collaboration, it's the policy I want, for a clean history and to avoid conflictors
17:38 Heffalump sm: it's the premise of your condition I object to :-)
17:39 Heffalump darcs doesn't work great with conflicts, but it's not that bad for minor ones
17:39 sm well, what would you do ? give me a simple answer :)
17:39 Heffalump I'd allow the conflicts unless they were major
17:39 Heffalump but I would like rebase to be integrated in darcsden, I just want to de-emphasise it
17:40 Heffalump and yes, rebase does generally require manual intervention. If it didn't, you probably shouldn't have rebased in the first place.
17:40 Heffalump The manual intervention is generally conflict resolution.
17:42 sm ok, say the conflicts are major
17:45 Heffalump sm: ok, so my starting point is "let's not make the 'rebase from upstream' button too obvious"
17:45 Heffalump but given that, it should look somewhat like your "merge from downstream" button
17:46 Heffalump but, we need a conflict resolution UI first. With that, rebasing should be relatively easy to integrate.
17:46 Heffalump and yes, we need more documentation, which is part of why rebase isn't ready for release
17:47 Heffalump anyway, so the concrete step is that we should figure out a conflict resolution UI for darcsden
17:56 sm Heffalump: actually a darcsden ui is more secondary, I was thinking what do you do at the command line ?
18:03 sm ie what is it Aditya needs to do to rebase a bunch of patches against HEAD if I make a bunch of global changes. Pre rebase, I think the answer was unrecord them all, pull -a, resolve conflicts, re-record his patches
18:04 sm Now, I think he can do.. darcs rebase pull, then (darcs unsuspend; resolve conflict; re-record) for each patch ?
18:09 Heffalump sm: that's right, yes
18:11 sm ok, that sounds not bad, and maybe answers my how to clean up local history as well
18:12 sm I will practice some more
18:13 sm I think this means I needn't ever worry about trunk changes inconveniencing downstream contributors
18:19 sm regarding "Everytime he did it, I had to mess around with my local checkouts of his work" - I guess if your upstream rebases, it means you also need to rebase
18:25 Heffalump sm: correct
18:25 Heffalump and darcs doesn't have any metadata to track it, so it's a mess
18:25 Heffalump I think that's true of git too, but I'm not sure
18:26 sm darcs at least remembers the patches and their original content for you. I think I'll have to try it myself to understand what's missing
18:28 sm something about remembering your per-hunk choices from previous rebases, is it ?
18:30 sm I see http://git-scm.com/2010/03/08/rerere.html
18:33 Heffalump I think rerere is basically a conflict resolutionc ache
18:33 Heffalump "given this conflict you saw before, here's how to resolve it again"
18:33 Heffalump with darcs that's not really the fundamental problem (I don't think)
19:52 int-e the hope would be that you commute each pair of patches only once and never again, right?
19:58 amgarchIn9 joined #darcs
20:04 Heffalump int-e: right
20:21 sm_ joined #darcs
20:23 mizu_no_oto joined #darcs
20:25 raichoo joined #darcs
21:16 mizu_no_oto joined #darcs
21:26 whaletechno joined #darcs
21:35 raichoo joined #darcs
23:07 preflex joined #darcs

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