Camelia, the Perl 6 bug

IRC log for #darcs, 2013-05-31

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

All times shown according to UTC.

Time Nick Message
00:55 konundra joined #darcs
01:40 intripoon joined #darcs
01:52 amgarching joined #darcs
03:04 mizu_no_oto joined #darcs
03:24 alexei joined #darcs
03:42 preflex_ joined #darcs
04:46 mtp joined #darcs
05:17 carter joined #darcs
05:18 carter anyone know how to get a password reset for a darcs den account?
05:19 dolio Step 1: mentor a summer of code project...
05:20 carter dolio well, good thing I am :)
05:20 dolio Not that one, though. :)
05:31 carter dolio are you going to be at hackphi? :)
05:31 dolio Yeah.
05:31 carter dolio realistically i'm not using my darcs den account so its ok
05:32 dolio I do think that's one of the items in one of the summer of code projects.
05:32 dolio Otherwise, sm can probably manually reset it.
05:33 carter ok
05:33 carter cool
05:48 lelit joined #darcs
06:10 Heffalump carter: darcsden.com would be alexsuraci, hub.darcs.net would be sm
06:11 carter Heffalump wait… they're different?
06:11 carter huh
06:11 Heffalump same software, different instances
06:11 Heffalump hub.darcs.net is probably the more actively used/maintained one nowadays
06:12 carter ohhhhh, i had a hub accound
06:12 carter thanks
06:12 carter :)
06:30 kmels joined #darcs
06:41 carter joined #darcs
07:45 _ilbot joined #darcs
07:45 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.8.4 is out http://darcs.net/Releases/2.8
07:52 donri joined #darcs
08:20 idnar joined #darcs
08:32 nomeata joined #darcs
09:03 owst joined #darcs
09:29 idnar joined #darcs
09:42 alexei_ joined #darcs
09:52 * owst remembers that he doesn't have the magic of the GHC type-nats branch yet:  Couldn't match type `2 + 1' with `3'
11:11 nomeata joined #darcs
11:17 mizu_no_oto joined #darcs
11:21 kaol joined #darcs
11:26 mndrix joined #darcs
12:25 mizu_no_oto joined #darcs
12:36 konundra joined #darcs
13:19 edwardk joined #darcs
13:33 uniquenick joined #darcs
14:02 raichoo joined #darcs
14:07 qinglingquan joined #darcs
14:08 qinglingquan left #darcs
14:44 edwardk joined #darcs
14:49 nomeata joined #darcs
15:03 bsrk joined #darcs
15:09 uniquenick joined #darcs
15:28 nomeata joined #darcs
15:38 javier_rooster joined #darcs
15:42 mizu_no_oto joined #darcs
16:03 dolio joined #darcs
16:09 dolio joined #darcs
17:04 konundra joined #darcs
17:17 nomeata joined #darcs
17:24 edwardk joined #darcs
17:39 * Heffalump appears, belatedly
17:39 Heffalump bsrk: you around? Sorry, my laptop is misbehaving, though it seems ok again for now
17:40 bsrk hi
17:40 bsrk I am around
17:41 bsrk I have finished login using github
17:41 bsrk try it at http://106.187.94.254:8900/login/github
17:41 bsrk note that you do not have an account now
17:41 bsrk (I deleted all data)
17:42 _ilbot joined #darcs
17:42 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.8.4 is out http://darcs.net/Releases/2.8
17:43 bsrk joined #darcs
17:43 bsrk And then you will be able to login again with no password
17:43 * Heffalump tries it
17:43 Heffalump so, what's the model for github login? You have to register first?
17:44 bsrk Yes
17:44 bsrk A github account is linked to atmost 1 darcsden account
17:44 Heffalump and can you associate a github login with an existing account?
17:44 bsrk That part is not done yet, it is in the works
17:45 bsrk I expect it will be done tommorrow
17:45 Heffalump ok, so you can (a) register a new account from your github one and (b) associate an existing account with a github login
17:45 Heffalump and in both cases the result is the same in that youhave an account you can login to directly or via github?
17:46 bsrk Yes
17:46 Heffalump ok, cool
17:47 Heffalump a couple of things occur to me: (a) users might not want to have to set a password when creating a new account from a github one
17:47 bsrk Yes, I will make the password optional if you are registering using a github account
17:47 Heffalump (b) is the ssh key copying a one-off or can you sync up again later?
17:48 bsrk Once we have a mechanism for syncing existing accounts, it will also allow resyncs
17:49 Heffalump ok
17:50 Heffalump your blogpost looks good
17:51 Heffalump so you'll keep on with the github stuff for the next few days?
17:51 bsrk Yes
17:51 Heffalump I saw your patch for exporting the packs code in darcs - I'll get that applied shortly
17:52 Heffalump what will the server do with packs? Will it rebuild them on each new patch arriving, or only every so often?
17:52 bsrk Packs has to be rebuilt manually by the user
17:53 Heffalump via some button in the UI?
17:53 bsrk Yes
17:53 bsrk It is in the respository edit page
17:53 Heffalump that seems a bit inconvenient - isn't it the kind of thing that should be managed automatically?
17:54 bsrk It is because packs is mostly useful for larger repos
17:54 Heffalump the "Packs enabled" banner when you do it is a bit misleading btw, I think it should say "Packs built" as "enabled" implies some kind of boolean state being changed rather than a one-off operation
17:54 Heffalump how is the user supposed to know that?
17:54 Heffalump and how do they decide what the threshold is?
17:55 bsrk Yes, I will change that.
17:55 bsrk I am not sure of the threshold myself. :-)
17:55 bsrk I would say that when get is becoming slow, they need to rebuild it.
17:57 Heffalump you could benchmark yourself to figure out the criteria for rebuilding packs
17:57 sm users shouldn't know about packs imho
17:57 Heffalump in terms of the cost to the server in CPU and disk space, and to the users of not rebuilding in slow gets
17:57 Heffalump and then either bake the logic into the code, or make it something that the darcsden administrator tunes (but still with good defaults)
17:58 bsrk Heffalump: that seems a good idea
17:58 Heffalump and yes, exactly what sm said; there's no reason they should know they exist at all
17:58 bsrk 1) Find out the threshold from where using packs is useful
17:58 bsrk 2) Find out how many patches makes get slow
17:59 bsrk And configure an automatic mechanism by these two parameters
17:59 bsrk Okay, this seems a good idea
18:00 Heffalump I suspect the natural threshold is in terms of either the number or the total size of patches that are not yet in a pack
18:00 bsrk Yes, I think so too
18:00 Heffalump but there may be other variables such as the total repo size that play a part
18:02 Heffalump where's your development repo itself?
18:02 bsrk It doesn't seem logically connected. I think that if get indeed go slow despite the automatic mechanism, the user should build packs manually?
18:02 bsrk http://hub.darcs.net/Aditya/darcsden-gsoc
18:03 Heffalump you should note that in your next blogpost (unless I didn't spot it in the current one)
18:03 bsrk Yes, I forgot to point that out.
18:03 Heffalump actually key URLs like the development server and your development repo should be in each blgopost, so they at least somewhat standalone
18:04 Heffalump so that if someone just sees a recent blogpost they still know enough to take a look without having to read the whole history
18:04 sm +1
18:04 bsrk I will do that.
18:07 konundra joined #darcs
18:08 Heffalump Anything else to discuss now? I'll try to take a look at the code before our next meeting. When would you like that to be?
18:08 bsrk How about monday?
18:08 Heffalump sure
18:09 Heffalump I'm not quite sure what's up with my laptop, so it's possible I will just fail to appear like today if it decides to misbehave again
18:09 carter joined #darcs
18:10 Heffalump it took me about 20 mins of poking it to get it working just now - it was fine just before I left work, then I unslept it and it locked up and wouldn't boot again.
18:11 bsrk That's computers for you. They fail without giving notice. :-)
18:12 Heffalump there has been something a bit odd with the hard drive or controller for a while, but t's always worked again after a reboot. I'm not quite sure what to do about it now...
18:13 Heffalump anyway, see you on Monday.
18:13 bsrk bye
18:42 mizu_no_oto joined #darcs
19:06 gh_ joined #darcs
19:36 carter joined #darcs
19:51 mizu_no_oto joined #darcs
20:02 tomejaguar joined #darcs
20:03 kmels joined #darcs
20:24 mizu_no_oto joined #darcs
20:33 nomeata joined #darcs
20:58 mizu_no_oto joined #darcs
20:59 lelit joined #darcs
21:07 alexei_ joined #darcs
21:16 adnap Hello
21:17 Heffalump hi
21:17 adnap I copied a darcs repo and made some changes...
21:17 adnap I haven't recorded the changes, but I want to e-mail the maintainer my changes
21:17 adnap There is still one compiler error caused by my changes that I want the maintainer to look at
21:18 adnap How should I send him the changes?
21:21 mtp darcs diff -u | mail -s "Tentative patch" maintainer@example.org
21:21 mtp ?
21:22 adnap Can you apply a diff with darcs?
21:23 adnap Also, it seems -u is the default
21:23 mtp well, you're explicitly asking "how do i avoid using darcs to send the changes" here
21:23 mtp so, no,, you'd use 'patch' to apply the diff
21:24 mtp i mean, if you want to use darcs, you can arrange with the maintainer to pull from your repo, correct your compiler error, then you can 'obliterate'
21:24 mtp (and he can amend-record)
21:27 Heffalump or just use darcs send to send the patch
21:27 Heffalump darcs record, darcs send, darcs unrecord
21:28 Heffalump obviously you need to communicate somehow that the patch is just to look at
21:28 mtp oh, unrecord was what i was thinking of, not obliterate
21:28 mtp mea culpa
21:30 adnap I just sent him a diff
21:30 adnap src/Hint/InterpreterT.hs:115:21: `ldInputs' is not a (visible) constructor field name
21:30 adnap src/Hint/InterpreterT.hs:116:34: Not in scope: `ldInputs'
21:30 adnap Oops
21:30 adnap (My keyboard just landed on my mouse)
21:31 carter joined #darcs
21:34 raichoo joined #darcs
21:34 Heffalump in general a darcs patch is better than a diff (if the recipient has darcs) because darcs knows how to apply it to a different context safely
21:40 Igloo joined #darcs
22:17 raichoo joined #darcs
22:28 dolio joined #darcs
22:43 nomeata joined #darcs
22:55 sm mtp: you could also push it to a branch on darcs hub and send him the patch url
22:55 mtp fair enough
23:07 sm and he can merge it with a click
23:20 Heffalump but that's not what you want if you are just investigating locally, is it?

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