Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2014-02-06

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

All times shown according to UTC.

Time Nick Message
00:01 mornfall joined #darcs
01:08 favonia joined #darcs
02:24 intripoon_ joined #darcs
02:28 dolio joined #darcs
02:46 mizu_no_oto joined #darcs
04:22 favonia joined #darcs
05:14 mizu_no_oto joined #darcs
05:46 rdesfo left #darcs
07:00 gal_bolle joined #darcs
07:05 lelit joined #darcs
08:12 raichoo joined #darcs
08:58 mornfall joined #darcs
09:00 raichoo1 joined #darcs
09:11 nomeata joined #darcs
13:19 gh_ joined #darcs
13:23 gh_ notdan, about http://irclog.perlgeek.de/darcs/2014-02-04#i_8232266 , I think it's good
13:52 colDrMcBeardman joined #darcs
14:17 mizu_no_oto joined #darcs
16:02 stepcut joined #darcs
16:18 raichoo joined #darcs
16:45 mizu_no_oto joined #darcs
17:00 mizu_no_oto joined #darcs
17:32 mizu_no_oto joined #darcs
17:33 rdesfo joined #darcs
17:52 Heffalump sm: have you used cabal sdist on darcsden lately? I think it was broken by a patch of yours in July that added all the extra-source-files, though I'm not entirely clear what's wrong
17:54 Heffalump oh, and darcs test --backoff erally rocks on a simple reproducible case like this
18:02 Heffalump ah, i found the problem on Linux - public/uptime.txt doesn't exist. But it's still broken on Windows, how confusing.
18:08 sm Heffalump: I haven't, thanks for looking at it
18:08 sm and +1 for mentioning darcs test.. uh hmm how is that related to darcs trackdown ?
18:09 Heffalump ah, no it does fix it on Windows, I was testing the wrong tree
18:09 Heffalump darcs trackdown got merged into darcs test a while ago
18:09 Heffalump hmm, perhaps only in 2.9
18:09 notdan darcs trackdown is older, isn't it?
18:10 sm I see.. no trackdown command (not even an alias) in darcs 2.9.8+21
18:10 Heffalump is it in 2.8 ? If so we should add at least an error message to 2.9
18:11 sm darcs test looks pretty cool and underused, I shall look for opportunities to use it
18:11 notdan http://darcs.net/manual/bigpage.html the official manual mentions trackdown
18:12 notdan oh the manual is for 2.8.1 only
18:12 sm yes, both trackdown and test are in 2.8.4
18:12 mizu_no_oto joined #darcs
18:13 notdan OK, that's weird. http://darcs.net/manual/ is a manual for 2.9.1, http://darcs.net/manual/bigpage.html is a manual for 2.8.1
18:13 notdan the darcs.net homepage mentions the latter
18:13 notdan additionally, there is a link to the 2.8.1. pdf manual
18:14 sm oh oh
18:14 sm doc rot
18:14 notdan should we generate a 2.8.4 manual?
18:14 sm yes, since it's the current release
18:15 sm ideally we'd keep manuals for the major recent releases online, like postgres
18:15 Heffalump ouch :-( I vaguely recall kowey knows about the setup, but I could be wrong.
18:15 notdan I could look around for a way to generatet a manual
18:16 Heffalump notdan: can you remind me where to look for the latest version of your whatsnew -i patch?
18:16 notdan but one thing I am confused about is the current darcs version
18:16 notdan Heffalump: http://hub.darcs.net/co-dan/darcs-screened
18:16 Heffalump I don't know why the latest devlopment version would be at the main URL
18:16 notdan So 2.9 is a development version?
18:16 notdan the whole 2.9.* series?
18:16 Heffalump notdan: yes
18:17 Heffalump notdan: is the import changes patch there still appropriate?
18:17 sm notdan: you can tell the current version by what's on hackage
18:18 notdan Heffalump: well, I guess you guys favoured a more compact style; I have not updated it to match this yet
18:19 notdan do you want me to do it now/tonight?
18:19 Heffalump sorry, no rush, I was just worried I was letting your patch languish unreviewed
18:20 notdan yeah I forgot I have to fix the imports
18:20 Heffalump actually if you like I'll sort that out with rebase when I review it
18:22 notdan OK that's fine
18:23 notdan syntactically those two patches do not depend on each other
18:23 Heffalump sm: one patch for you in ganesh/darcsden
18:23 notdan so we can work concurrently :)
18:23 Heffalump oh, right - but they do depend semantically, don't they? As you use new imports in the code patch.
18:24 sm great, thanks
18:24 * Heffalump is just getting home (funny timings today due to a tube strike), back online later
18:24 notdan Heffalump: yep
18:24 notdan Heffalump: by the way, by 'review' you mean 'get into -screened' or 'get inro -reviewed'?
18:25 Heffalump is it ok to get the addition of the imports into the same patch as the code changes, or at least a separate one to the imoprt reformattnig?
18:25 Heffalump <gone>
18:25 notdan cya soon
18:26 notdan Now that you  mentions is, it might be sensible to put the addition of imports/exports in the same patch as the actual implementation
18:26 notdan just to make it self-contained
18:26 notdan By the way, one change I would strongly advocate for getting into -reviewed (and even possibly backported for 2.8) is this: http://hub.darcs.net/darcs/darcs-screened/patch/20140120220552-e7f45
18:27 notdan It is a nasty bug and people still encounter it every time they want to install darcs
18:27 sm +1
18:33 sm darcs.net setup/admin remains a bit of a bottleneck for us I think
18:33 sm I wonder what's needed to get darcs 2.10 out
18:34 notdan ok , darcs screened does not have a manual...
18:35 notdan at least I cant find one
18:35 sm there was a big change in the manual, it was at least partly moved from inline literate haskell to separate markdown docs I believe
18:36 notdan Fri Jul  6 00:57:12 MSK 2012  Guillaume Hoffmann <guillaumh@gmail.com> * remove website and manual
18:36 notdan sm: in 2.8 the manual is a bunch of TeX files
18:36 notdan that were compiled to html
18:36 sm I think those were also moved to markdown
18:38 notdan there are no markdown files in the repostiroy
18:41 notdan well anyway, I think gh knows how to build the docs
18:49 Heffalump sm: fix packs, finish rebase to an acceptable standard
18:49 Heffalump not sure what else
18:50 sm this manual situation
18:50 sm release notes
18:51 sm about rebase: could you provide pull --rebase, that just works ? That would be huge
18:51 sm no ui needed
18:51 lelit joined #darcs
18:55 mornfall joined #darcs
18:59 Heffalump darcs rebase pull does exist, but may not do what you want..
19:02 sm I mean darcs pull --rebase that does the same as git (amends any unpushed local commits so they apply cleanly on the latest from upstream)
19:10 notdan OK, darcs 2.8.4 manual: http://covariant.me/stuff/darcs/darcs_28_manual/ PDF: http://covariant.me/stuff/darcs/darcs_28_manual/darcs.pdf
19:10 notdan Archive: http://covariant.me/stuff/darcs/darcs_28_manual/manual.tgz
19:10 notdan this should probably go in place of http://darcs.net/manual
19:27 sm great
19:27 sm the wiki front page should be updated, I forget how
19:30 notdan maybe a patch should be sent to http://darcs.net/darcs-wiki?
19:32 sm yes, that or someone here with write access to darcs.net:/var/lib/gitit/wikidata/FrontPage.page can edit that
19:32 notdan Do you guys wanna do another Darcs (Weekly) News issue?
19:32 sm *Weekly* ? really ?
19:33 notdan Erm yeah
19:33 notdan just news :)
19:33 sm weekly would be awesome :)
19:33 notdan sorry I mixed up haskell weekly news
19:34 sm hmm we also have the wiki mirrored at http://hub.darcs.net/darcs/darcs-wiki
19:36 sm so many tools, so hard to fix a link :/
19:39 notdan heh :)
19:39 notdan there has to be a person with access
19:39 notdan I think commiters should also have access to darcs-wiki
19:39 sm Heffalump, kowey, gh afaik
19:41 Heffalump I have access to darcs-wiki
19:42 sm the files are writable by the wiki-author group which has Heffalump, markstos, and gitit as members
19:42 Heffalump I thought any darcs committer does have wiki access, but in practice it may not be properly configured everywhere
19:42 Heffalump certainly in theory they should
19:48 sm I note darcs.net is running debian lenny which was end-of-lifed (no more security updates) two years ago
19:49 sm don't let me depress you, just sharing the info while it's in front of me
19:50 notdan :P
19:51 xymox joined #darcs
19:51 sm I remember kowey threatening to upgrade this machine a while back
19:55 sm back to the wiki.. FrontPage, Thanks and Donations could be edited like other pages if they were removed from no-edit list at /etc/gitit.conf
19:56 Heffalump well, I guess that's deliberate. Not sure if the paranoia is justified.
19:56 sm then we'd have to be more vigilant about spam edits to the front page
19:57 Heffalump yeah
19:57 Heffalump btw, you can darcs send patches to the wiki. Also I'd be happy to give either of you write access to it if you want.
19:58 sm it doesn't look like gitit can send edit notifications by mail
19:59 xymox joined #darcs
20:01 notdan Heffalump: can you add me to the wiki-author group?
20:01 notdan I promise not to spam :P My username on the wiki is 'Dan'
20:01 Heffalump I think you have to have a proper shell account on darcs.net
20:02 sm yes you do.. gitit (at least this version) doesn't have groups
20:02 notdan oh :(
20:02 sm we've handed out quite a few shell accounts over the years
20:02 Heffalump that's fine, I can make one
20:02 sm want one ?
20:03 notdan If that's alright with you
20:03 sm +1
20:03 Heffalump pick a username and let me have a ssh public key (email ganesh@earth.li)
20:05 notdan done :)
20:06 Heffalump bear with me while I figure out how things work again
20:10 mizu_no_oto joined #darcs
20:11 Heffalump notdan: ok, see if you can ssh in and if you seem to be in the wiki-author group
20:12 Heffalump if so you should be able to push patches to /var/lib/gitit/wikidata - best to make a symlink in your home directory
20:13 gh_ joined #darcs
20:13 notdan Permission denied (publickey).
20:14 sm .ssh and below owned and readable only by dan ?
20:14 gh_ hi
20:15 Heffalump ah, that's what ! means in the shadow entry :-) ("account is locked") Try now.
20:15 Heffalump hi gh_
20:15 sm hi gh_
20:15 gh_ notdan, sm, about the manual and current/2.10 darcs, yes there is a documentation rot problem, sorry about that
20:15 Heffalump gh_: were your changes to hashed-storage as you imoprted it (a) changing imports and (b) turning a doc comment on types into doc comments on its fields?
20:15 Heffalump sm: I thought of that one in advance :-)
20:15 gh_ in http://darcs.net/Releases/2.10 I wrote that we should fix it for 2.10
20:16 gh_ Heffalump, yes
20:16 gh_ Heffalump, nothing more really, no "real code" change
20:16 Heffalump yep, thanks, just checking I'd worked out the right baseline
20:17 Heffalump you also added a file called 'review' :-)
20:17 gh_ hmm?
20:17 gh_ oops!
20:18 gh_ sorry, that was not meant to be included
20:18 Heffalump I didn't think it was
20:19 notdan OK ssh works now :) thanks
20:19 sm Heffalump: you could add simonmichael to wiki-author while you're at it
20:20 notdan gh_: I built manual for 2.8.4: http://covariant.me/stuff/darcs/darcs_28_manual/manual.tgz
20:20 notdan but I don't know how to build the manual for 2.9
20:21 Heffalump sm: done
20:22 gh_ notdan, thanks, I think we should publish the bigpage manual of 2.8.4, and nothing else until 2.10 goes out
20:23 gh_ because I think the markdown documentation we will generate for 2.10 will also be a single document (html, pdf, etc.)
20:24 notdan Fair enough, but could you document the process of generating the manual anyway?
20:24 sm thanks. There was one small junk change pending on the Troubleshooting page, I've reverted it
20:24 notdan the bigpage.html is http://covariant.me/stuff/darcs/darcs_28_manual/bigpage.html btw
20:24 notdan I wonder where should I put it on the server
20:25 sm notdan: darcs.net is served out of /home/darcs-unstable IIRC
20:25 gh_ notdan, yes we have to do it. it's not yes documented but a mix of some wiki pages (Features GettingStarted etc.) and the output of `darcs help markdown`
20:25 sm /home/darcs-unstable/website
20:26 sm gh_: is the wiki the master copy of part of the manual now ?
20:28 notdan sm: thanks
20:28 gh_ notdan, do you have access to the darcs.net server? otherwise I can upload the manual myself.
20:29 gh_ sm, sort of but I/we haven't started doing anything serious about it
20:29 sm is that the plan ?
20:29 notdan gh_: I do have access, but not to the website directory :)
20:29 notdan If you can download http://covariant.me/stuff/darcs/darcs_28_manual/bigpage.html and http://covariant.me/stuff/darcs/darcs_28_manual/darcs.pdf to there it would be great
20:31 gh_ notdan, ok, I'm looking at it now
20:38 gh_ notdan, done, strangely the path was ~/reviewed/doc/manual ... not super safe
20:38 gh_ ~/website/manual is not taken into account
20:39 gh_ I guess that relies on some apache configuration somewhere
20:39 sm we may want website/manual/X.Y/...
20:40 gh_ yes
20:41 sm and you're right, there's a RewriteRule ^/manual/(.*)$ /home/darcs-unstable/reviewed/doc/manual/$1
20:41 sm in /etc/apache2/sites-enabled/000-default
20:42 notdan maybe we can delete website/manual then?
20:42 sm I hope it's ok to talk about paths etc. on this logged channel, I feel we need to
20:44 sm I wonder why the document root isn't website/ instead of having a rewrite rule for everything
20:44 gh_ notdan, yes I've deleted it
20:44 gh_ sm, I guess the idea was for the manual to be updated along with the reviewed repository
20:44 sm oh, so that you can darcs get http://darcs.net I guess. I think that's an anti-feature now
20:46 gh_ `darcs get http://darcs.net` is an anti-feature?
20:47 sm yeah, it doesn't make much sense that you'd get darcs source that way; I'd expect to get the website, or nothing
20:47 sm and you don't know which branch you're getting
20:47 sm and it complicates our apache config
20:47 Heffalump I think we thought people would assume it was 'HEAD'
20:47 Heffalump whatever that is in our own nomenclature
20:48 sm ah is it time for me to moan about -reviewed and -screened again ? oh say yes
20:50 sm darn
20:50 notdan I have never heard your rant, sm :)
20:51 Heffalump and what should we call them? (I may have forgotten this answer from your last rant, which I probably have heard :-)
20:51 sm I have condensed it, through practice :)
20:52 sm our branches are confusing and more overhead than we need at our current size/activity level. We should have darcs trunk, and release branches (and sundry contributor branches)
20:53 sm in conclusion: we should simplify. Thank you
20:53 Heffalump so which bit of the current process should we omit? screening (quick glance at the patch to make sure it's sane and won't make a total mess of history if it gets undone later, or reviewing?
20:55 Heffalump (or change it to somehing else, I tguess)
20:55 sm I think screening - preliminary review - should be done in contributor repos, before merging to trunk.
20:56 Heffalump I'm not quite sure I see the distinction
20:56 sm and any extra pre-release review/fixups can be done any time before release, in trunk or in the release branch
20:58 sm Heffalump: you don't see what I'm suggesting as being any simpler ?
20:58 Heffalump I'm not quite sure I've understoodd it
20:59 Heffalump anyway, I think the first thing we should do (though it's not trivial) is migrate to hub
20:59 Heffalump then we can play with the workflows there
21:01 sm I'm saying we shouldn't have two separate official darcs repos of equal prominence; it's unusual and makes the contribution process non-obvious
21:02 sm we should have a single unambiguous trunk
21:02 Heffalump the -reviewed branch is really only for the review team - we could make it less prominent
21:03 sm there is no review team, until (if we're lucky) release time
21:03 sm I don't think that should exist as a long-lived branch
21:04 Heffalump are you saying there's no review team in practice now, or that there wouldn't be one in your new process?
21:04 sm the first
21:04 Heffalump s/review team/committers/ if you like, anyway
21:05 sm currently, our review process is more organic, we chat here and on the list and review each other's repos on hub
21:05 sm github style
21:09 sm does any of this resonate with anyone ? It seems like an easy win to me, but I don't seem to be communicating that
21:10 sm aside from inertia, I don't see how our two branches are helping us at all for a few years
21:10 Heffalump I still don't quite follow what scrutiny patches would get at what point
21:11 sm patches are reviewed in contributor repos before someone with the darcs commit bit merges to trunk. Like I review your darcsden patches
21:11 Heffalump ok, so the reason we brought in the two phases was that wasn't working when reviews didn't happen for many weeks
21:11 Heffalump because conflicts built up
21:12 sm ok, I think we have better solutions to that now
21:12 sm two phases hasn't solved it
21:12 sm (also)
21:12 byorgey_ joined #darcs
21:13 Heffalump I think it has made it better. We don't drop so many contributions on the floor
21:13 sm if a contribution doesn't get merged reasonably quickly, it probably isn't worth worrying about; or, we expect the contributor to rebase and republish their patches as many times as needed for review
21:14 gh_ I do reviews.. :(
21:15 gh_ not very deep ones, arguably
21:15 gh_ migrating to hub sounds interesting anyway
21:15 sm agreed on that
21:16 sm I also do reviews, when the code is viewable on hub
21:16 sm I mean, I'll look and squawk if I see something awful
21:17 gh_ and I also think the screened/reviewed distinction was an improvement when it happened
21:19 gh_ but I see what you're proposing
21:19 sm conceptually, -screened is the repo that's ok to obliterate/amend in, is that right ?
21:20 Heffalump I think if we made reviewed less visible, we could do at least some of what you're suggesting
21:20 Heffalump no, it's not ok there
21:20 gh_ I don't know how things work on github/bitbucket, but I'd like to be able to create a fork of HEAD on hub.darcs.net, from the commandine, and then push my patch there
21:20 Heffalump gh_: bundles!
21:21 Heffalump forks make sense for longer-running work
21:21 Heffalump but darcs send should just work
21:21 gh_ Heffalump, bundles sometimes suck when they have private patches in their context
21:22 Heffalump shouldn't happen with darcs send, should it?
21:22 Heffalump not in the simple case
21:22 gh_ Heffalump, but maybe it should be one's responsibility to always work from HEAD + only the patches you want to send
21:23 Heffalump if you do darcs send http://darcs.net, darcs won't include your private patches in a context
21:23 Heffalump it'll either force you to include them, or leave them out entirely
21:23 Heffalump unless I'm truly confused
21:23 gh_ o_O
21:23 gh_ I'm not sure anymore
21:25 gh_ Heffalump, no, it doesn't force you to include them
21:25 Heffalump it would if you depended on them
21:25 gh_ Heffalump, only if there's some dependency going on
21:25 gh_ yes
21:25 Heffalump that what was I meant by the 'either'
21:26 gh_ oh wait, that 2nd patch was not included in the context
21:26 gh_ ok, I think what happened was cases when I tried bundles destined to HEAD, to by reviewed branch
21:26 Heffalump yes, that does complicate things
21:26 gh_ s/tried/tried to apply/
21:27 Heffalump hub can make it all simpler because it has darcs available to it (in the code)
21:27 Heffalump (with more dev work, naturally!)
21:27 * notdan is heading off to bed
21:27 sm gh_: ssh hub.darcs.net fork darcs/darcs; darcs get hub.darcs.net:darcs; darcs push
21:27 Heffalump notdan: night. I didn't get as far as doing anything with your patches after all, but will try again tomorrow..
21:28 sm (no, that doesn't work today)
21:28 sm I could be wrong but I feel like the answer to "how do I contribute" should almost never be "darcs send" these days
21:28 Heffalump darcs send which will upload patches by http to a suitable server
21:29 sm ok, but it's still confusing, I even forgot that it does that, and not enough people use it to rely on
21:30 Heffalump we need a suitable server first - which is hub once I/we sort out the potential issues with spam. What else could we alias it to to make it clear? darcs sumit?
21:30 Heffalump submit
21:31 sm send always sent by mail, I think it's always going to be confusing that it will do http under some circumstances
21:31 sm and we have darcs push
21:32 Heffalump so you think darcs push should turn into a submit if you don't have permission (e.g. to a http url)?
21:33 sm sorry if I'm forgetting all the valid reasons why folks worked hard on this, I'm thinking of the basic how do folks contribute to darcs workflow and I don't see what's wrong with the github-style fork & pull request flow
21:33 Heffalump I think I've ranted about it several times :-) But briefly, it's overly verose and it leaves litter all over the place.
21:34 Heffalump You could improve it with bundles even for git. But for darcs where patches are the first class objects, it's really silly.
21:38 sm ok. but we should support the familiar fork & pr flow that already works today, even while working on a more efficient alternative
21:38 sm and yes, having this feature under darcs push rather than darcs send does sound better IMHO
21:41 Heffalump I don't object to supporting that workflow, though I don't plan to work on it :-)
21:43 sm notdan: good night, and thanks for asking :)
21:44 sm I should go eat, but Heffalump re moving to hub ? how would that look
21:44 Heffalump well, moving the repos is trivial, so what we need to replace is the issue tracker and the patch tracker
21:45 sm do those need to happen all together ?
21:45 Heffalump the patch tracker, i.e. having a place where submissions don't get lost, is what I see as crucial
21:45 Heffalump no, I don't think so
21:46 Heffalump but I haven't worked out a detailed plan or anything. I can see pieces I want to make work (bundles in particular) before getting on to that
21:46 sm it seems to me we can move the repos any time without much disruption
21:47 Heffalump agreed, though I'm a bit reluctant to do that first for fear of encouraging contributions that way that we aren't ready for
21:47 sm and we have potentially three places where submissions don't get lost: the current patch tracker, the send/push-enabled darcs issue tracker on hub if we want, and contributor/feature branches  on hub
21:47 Heffalump that's three places where submissions _can_ get lost :-)
21:47 Heffalump by which I mean forgotton/overlooked, as opposed to destroyed
21:47 sm I don't think submissions getting lost as a current problem
21:48 Heffalump I do, we forgot some from Mark Stosberg that he submitted via hub
21:48 sm they're still there, showing up in the branches page
21:48 Heffalump yes, but noone actually looked at them
21:48 sm that'll change if we move to hub
21:49 sm I'm in favour of anything that can simplify and focus our attention in the right places
21:50 sm (I looked at them)
21:50 Heffalump well, the problems I see with the current workflow is you can't tell what's a submission and what isn't.
21:50 Heffalump for example darcsden is still cluttered up with Aditya's old gsoc branch
21:50 sm that is a problem, but not a big one I think. We can easily coordinate that
21:51 Heffalump how? The nice thing about the current patch tracker is that the patch statuses allow us to pick up again fairly easily after a period of inaction
21:53 sm issue pages can link to repos/patches of interest
21:53 Heffalump how do you easily get a list of "this is what I should look at"?
21:53 Heffalump as a reviewer
21:54 sm as soon as it becomes a problem, we will enhance darcsden in that area
21:54 sm there's no motivation to work on that because darcs doesn't use hub yet
21:54 sm and it hasn't bothered me sufficiently in the darcsden repo
21:55 sm for now the adhoc process would be what I've done with darcsden: "hey X, can you remove those obsolete repos/patches so they don't clutter my branches view"
21:58 Heffalump and where would we store discussion on patches?
21:59 sm on the relevant issue page, on the list, in this channel
22:01 Heffalump it feels like there'd be a lot of manual admin to do that we don't have with the current tracker
22:01 Heffalump I'd rather have substitutions in place first - not perfect, but better than now
22:01 sm well I'm not arguing to give up the current tracker
22:01 sm I haven't thought about the tracker too hard
22:02 Heffalump I really don't want to have to look in two places for submissions that need attention
22:02 sm ok, are there small steps we could take that don't create that situation
22:03 Heffalump well, I see making bundles work as a relatively small step that would be going in the right direction
22:03 Heffalump I'm actually working towards that in an oblique kind of way
22:03 sm cool. That requires coding and time, I'm sure we could do less
22:04 sm like, just for example, move the official repos but keep the official process pretty much as is
22:04 Heffalump what's the benefit in that?
22:07 sm it simplifies the situatino; it makes the official darcs repos browsable and easier to fork; it's a step forward in our goal of moving to hub - taking it will help the next step to become clear
22:07 sm what benefits are you thinking of in moving to hub ?
22:08 Heffalump having the workflow integrated in hub - online review, merging, browsing of submissions etc.
22:08 Heffalump the huge weakness of the current tracker is that the server doesn't have code written against the darcs library
22:09 Heffalump I have a local script that scrapes the patches from the server to construct a table of "what needs doing". That kind of thing could be integrated into hub (suitably generalised)
22:09 Heffalump or in the short/medium term via a darcs-repo-specific hack if you were amenable
22:11 sm I think having the darcs repos officially on hub will encourage more online review, also more experimentation with new processes and more work on darcsden
22:12 sm darcs-repo-specific hacks are totally fine
22:13 Heffalump I guess I don't see much difference between the repos being officially there and the current mirror
22:13 sm supporting darcs development is actually hub's #1 mission (though I moved it to #2 in the faq)
22:14 kowey joined #darcs
22:14 Heffalump but I wouldn't mind moving them there especially if you wouldn't mind keeping an eye on potentialy submissions that came in that way
22:14 Heffalump I guess there is a lot of minor housekeeping needed to make that move that we could shake out by actually doing it.
22:15 sm that (glancing at branches page now and then) is easy, happy to do it, but also not necessary if we haven't changed the contribution process yet
22:16 Heffalump ok, I guess next step is email darcs-devel saying you+I want to do it
22:16 sm another variant is to move hub to the darcs.net server. I'm not opposed to that, but it's got some downsides
22:17 Heffalump is the darcs.net server beefy enough?
22:17 Heffalump I guess using the present hub server leaves us a bit exposed if you're away
22:17 sm it looks like 512M. possibly not
22:17 sm it would run it, but maybe not build it
22:18 sm yea this is the kind of thing we should figure out
22:18 alexei joined #darcs
22:18 sm you mean if hub went down and I'm unreachable.. darcs is off-net
22:18 Heffalump yes
22:19 Heffalump or just a stuck repo as happens every so often
22:19 sm like we've had before with the wiki eg.. but we have several people with logins there
22:21 sm I wouldn't expect it to be a problem in the near to medium term, but there should be a plan for it
22:21 sm I could give you an account, eg
22:21 Heffalump you never go away without internet access? :-)
22:21 Heffalump yeah, that would help
22:22 sm not for too long, and hub has been pretty stable, and we all have local copies of the repos, but yes it could happen
22:22 sm hub also has a good and recently tested backup/restore policy, not sure about osuosl's
22:22 Heffalump good point
22:23 sm http://hub.darcs.net/#what-is-the-backup-policy for reference
22:23 kowey they just had some AC problem today, I think
22:24 sm kowey! did you sense a disturbance in the force ??
22:24 Heffalump I disturbed it :-)
22:24 sm nice
22:25 Heffalump not intentionally. darcs.net has monitoring for config changes
22:26 sm the most ideal imho would be to replicate my setup on a new linode vps just for darcs, but that seems to require money. Using mine or (with more effort) upgrading and running on osuosl's gets around that
22:26 * kowey unfortunately did things in the wrong order
22:27 sm are we a member of the SFC by the way ?
22:27 kowey and sent an uh-oh mail to support@ before thinking “oh, I know! I'll go on IRC!”
22:27 kowey sm: why yes
22:27 Heffalump how much money would we need for a darcs-only vps?
22:27 sm they could own a linode for us.. I don't suppose they'd pay for it too ?
22:28 Heffalump darcs has some money stashed with them
22:28 sm it's $20/mo for a 1G linode
22:28 Heffalump is that the spec of the current one?
22:28 kowey not something we could ask the osuosl folks about?
22:28 sm but 2G is a little better for hub, that's $40
22:29 Heffalump asking osuosl is certainly another option
22:29 sm it's currently on a 2G, but sharing with my other stuff
22:30 sm ask them what ?
22:31 sm for a newer & beefier vps from them ? That would be good, but not as good as a linode vps in a few ways
22:32 sm but better in other ways
22:32 sm I don't know that I'd ever trust their backups as much as linodes
22:32 Heffalump btw Hetzner seem to offer 2GB for 20euros (http://www.hetzner.de/en/hosting/produkte_vserver/vq19). Ah, dunno about backups though.
22:39 xymox joined #darcs
22:46 sm more to be revealed ?
22:46 sm thanks for the chat Heffalump
22:47 kowey good night, all! hmm, maybe hang out here tomorrow (admin catch up day)
22:47 sm (& all). Sounds good kowey. afk a bit o/
22:49 xymox joined #darcs
22:58 raichoo joined #darcs
23:01 xymox joined #darcs
23:13 xymox joined #darcs
23:22 xymox joined #darcs
23:57 mizu_no_oto joined #darcs

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