← Previous day | Index | Channel Index | Today | Next day → | Search | Google Search | Plain-Text | plain, newest first
All times shown according to UTC.
| Time | Nick | Message |
|---|---|---|
| 00:36 | twb joined #darcs | |
| 01:02 | lambdabot joined #darcs | |
| 01:40 | mjrosenb joined #darcs | |
| 01:41 | mjrosenb | is there any way to reverse the actions of a patch in the working copy, without changing the repository? |
| 02:19 | twb | There are no patches in the working tree. |
| 02:19 | I'm not sure I understand what you're trying to do. | |
| 02:20 | Do you mean something like hg update, where the working tree reflects an "old" state of the repository, i.e. without some of the patches applied? | |
| 02:20 | mjrosenb | yes |
| 02:20 | but i don't want to revert to an old state | |
| 02:20 | i just want to remove the effects of one patch | |
| 02:22 | twb | You can use "darcs rollback" to record a patch that does the opposite of another patch |
| 02:23 | Or you can "darcs get" the repo into another place, then delete that patch from that new repo | |
| 02:24 | mjrosenb | i don't want to change the repo, only the working copy |
| 02:29 | zooko | The idiomatic darcs workflow is 'make another repo'. |
| 02:30 | 90% of the answers to "How do I do X?" in darcs begin with "First, you make another repo". | |
| 02:30 | So, the way I would do it (although this is certainly not the only way) | |
| 02:30 | is make one repo to be "the one with all the patches" | |
| 02:30 | and another to be "the one with that one patch undone" | |
| 02:31 | mjrosenb | this is slightly problematic |
| 02:32 | hrmm | |
| 02:32 | zooko | Okay, if that doesn't work for you there are plenty of other ways to go about it. |
| 02:32 | mjrosenb | it might work |
| 02:32 | the main problem is | |
| 02:33 | that i have a bunch of local changes in a darcs repo | |
| 02:33 | zooko | I suspect that you have an idea of "the repo" as a canonical, central, append-only thing that you change only rarely or carefully. |
| 02:33 | mjrosenb | but i need to push back to an svn repo |
| 02:33 | zooko | Which is fine, but probably the darcs repo which is integrated into the local directory that you are working on should not be *that* repo. |
| 02:33 | mjrosenb | the svn repo has too many files to use darcs with |
| 02:33 | zooko | Yeah, and you want to svn commit without the effects of "that patch" being involved. |
| 02:34 | mjrosenb | indeed |
| 02:34 | zooko | What I would do is "move that patch aside", then svn commit, then "put it back". |
| 02:34 | The way to "move that patch aside" is make a new darcs repo which has that patch, then "darcs obliterate" that patch from your currnet repo. | |
| 02:34 | The way to "put it back" is darcs pull from the aside repo then /bin/rm -rf the aside repo. | |
| 02:35 | This sort of workflow is so common to me that I almost wonder if future versions of darcs should somehow make it more apparent to users. | |
| 02:35 | mjrosenb | but darcs isn't managing all of the files that svn is |
| 02:35 | zooko | So? |
| 02:35 | By the way, how many files are we talking about that are too many for darcs? | |
| 02:35 | And is the SVN repo public so that I can run tailor on it to produce a darcs mirror of it? | |
| 02:35 | mjrosenb | so if i clone the darcs repo, then i will not be able to commit back due to lack of svn metadata |
| 02:36 | zooko | You'll commit from the current darcs repo, not from the new one. |
| 02:36 | You'll svn commit, that is. | |
| 02:36 | cd .. && darcs get $REPO aside-repo | |
| 02:36 | mjrosenb | oh |
| 02:36 | o read that backwards | |
| 02:36 | zooko | cd $REPO && darcs oblit -p"$THATPATCH" |
| 02:36 | svn commit | |
| 02:36 | darcs pull ../aside-repo | |
| 02:36 | /bin/rm -rf ../aside-prepo | |
| 02:36 | /bin/rm -rf ../aside-repo | |
| 02:37 | mjrosenb | tailor? |
| 02:37 | 531950 files | |
| 02:37 | zooko | Wow! |
| 02:37 | That's a lot of files. | |
| 02:37 | Is it open source? | |
| 02:37 | mjrosenb | indeed |
| 02:37 | no | |
| 02:40 | oh shi~ | |
| 02:40 | i need to use plan b in any cose | |
| 02:40 | *case | |
| 02:42 | zooko | Ok. |
| 02:42 | mjrosenb | yeah, so upstream, there are a bunch of conflicts |
| 02:42 | so i checked out a new repo | |
| 02:42 | i'm going to darcs init + add all relevant files | |
| 02:43 | then push individual patches over | |
| 02:43 | zooko | It will probably not work. |
| 02:43 | mjrosenb | it may |
| 02:43 | due to the nature of the changes and the conflicts | |
| 02:44 | and i consider it easier than the alternitive | |
| 02:44 | zooko | Are you using darcs-2? |
| 02:44 | mjrosenb | 2.0.2 (release) |
| 02:45 | zooko | Hm. |
| 02:45 | I don't think your plan can work. | |
| 02:45 | You'll be creating,in the new repository, a patch that says "create this file named $FILE and add these complete contents to it". | |
| 02:45 | Then you'll ask darcs to pull in a patch from the old repository which says "Make this change to this file". | |
| 02:46 | Darcs will either lock up entirely or else just treat the whole thing as a merge conflict and ask you to sort out what the new contents of $FILE should be. | |
| 02:47 | mjrosenb | why would it lock up entirely? |
| 02:48 | twb | 13:25 <mjrosenb> i don't want to change the repo, only the working copy |
| 02:48 | mjrosenb: unlike most other dVCSs, Darcs ties the working tree and the patch set together tightly | |
| 02:49 | This means we don't have a lot of confusing commands to separate the two, and you can never get in a consuing (to new users, at least) state where you've downloaded patches into the repo, but the working tree is not updated | |
| 02:49 | The downside is, of course, that you need to make temporary repos (branches) to do some stuff that other dVCSs can do in a single more complicated style of repo | |
| 02:49 | mjrosenb | fair enough |
| 02:50 | zooko | It might lock up entirely because it can't figure out how to merge a patch which created a file named FOO with XYZ contents in this directory |
| 02:50 | I mean in this repo, | |
| 02:50 | with a patch which created a file named FOO with XYZ, or possibly with XYQ contents in this other repo. | |
| 02:50 | Or maybe it won't. I haven't tried that in years. | |
| 02:51 | It would definitely have locked up in, uuh, 2005 or so. | |
| 02:51 | Maybe try it on just one file before putting a lot of time into it? | |
| 02:58 | kpreid_ joined #darcs | |
| 03:06 | Jerri joined #darcs | |
| 03:35 | mjrosenb | is there an easy way to view a single patch? |
| 03:36 | gwern | darcs changes -i? |
| 03:36 | mjrosenb | indeed. thanks |
| 03:41 | semka joined #darcs | |
| 03:42 | twb | Darcs diff --patch foo also is nice |
| 03:42 | semka | good morning! |
| 03:42 | by the way, is there gitk analog for darcs? | |
| 03:42 | mjrosenb | ha-hah, --reverse makes it run in a reasonable order |
| 03:43 | e_s_p joined #darcs | |
| 03:43 | e_s_p | Hi all |
| 03:43 | Question: I'm a bad person. | |
| 03:43 | I've duplicated tags in a repo a couple of times | |
| 03:44 | Like, I tag something version 0.3.2, but then I realize I didn't increment the version number in the main code, so I do that, then I tag it 0.3.2 again. | |
| 03:45 | My question: how to get rid of the first tag? | |
| 03:50 | twb | semka: only in Emacs |
| 03:50 | semka | twb: that's nice! |
| 03:50 | twb: where i can read more about this? | |
| 03:50 | twb | Ask fsbot about darcsum, dvc and vc-darcs. |
| 03:50 | They aren't very good | |
| 03:51 | semka | thanks |
| 04:38 | twb | I'd just like to reiterate that hg's merging is FUCKING RIDICULOUS and I am IMMEASURABLY thankful that I don't have to deal with it at home. |
| 05:06 | semka joined #darcs | |
| 05:28 | twb | Igloo: congrats on the camp vs. git video. The key point that I'd always missed was the a "visualize" equivalent for darcs/camp -- it makes it a lot clearer to the ignorant masses, I think. |
| 05:31 | semka | video is impressive, yes |
| 05:36 | twb | Actually that viz tool would be useful for me in darcs, but I suppose I'd need to use camp to get it :-( |
| 06:00 | lambdabot joined #darcs | |
| 06:31 | gour joined #darcs | |
| 06:39 | sjanssen joined #darcs | |
| 06:46 | semka joined #darcs | |
| 07:22 | _keturn joined #darcs | |
| 08:01 | balor joined #darcs | |
| 08:26 | C-Keen | Repository: /Users/klm/Projects/DAB-TI/dmb-dec |
| 08:26 | Kind: hg | |
| 08:26 | I have never seen this before | |
| 08:26 | what does determine the 'Kind:' | |
| 08:33 | twb | _darcs/repository/format, I think |
| 08:33 | Or something like that | |
| 08:33 | _darcs/format | |
| 08:38 | semka | is there any rss-feed or anything with recent "commits" to darcs.net repo? |
| 08:38 | lelit | C-Keen: where are you seeing that? it's very similar to the first patch made by tailor... :) |
| 08:40 | C-Keen | lelit: it is indeed a converted repo made by tailor |
| 08:42 | lelit | then it's the original repo format |
| 08:42 | C-Keen | ah ok |
| 08:42 | isn't this a bit confusing | |
| 08:44 | lelit | you're the very first person to raise the issue: the complete patch log seems clear enough to me... doesn't it says "this is the tailorized version of..."? |
| 08:49 | C-Keen | it says what I pasted |
| 08:51 | semka: it's a bit hidden: http://allmydata.org/trac/darc[…]n_copy&format=rss (and out of date) | |
| 08:51 | lambdabot | http://tinyurl.com/5ouxj9 |
| 08:51 | semka | yep, i already found this (: |
| 08:51 | but anyway thanks | |
| 08:52 | C-Keen | semka: adding a darcs2rss would be nice indeed |
| 08:53 | lelit | C-Keen: strange, the typical message is like this: http://pastebin.com/d76909376 |
| 08:55 | C-Keen | I have used tailor 0.9.35 |
| 09:06 | gour | how do you like darcs-fast-export? |
| 09:06 | lelit | C-Keen: it's been so from the very beginning |
| 09:06 | mjc joined #darcs | |
| 09:06 | gour | lelit: have you helped miklos with it? he gives you some credit in docs |
| 09:07 | mjc left #darcs | |
| 09:08 | lelit | gour: no, I'm focused on completely different stuff, and trying to find a new job |
| 09:09 | whenever I get there, I'd like to think about integrating the functionality in tailor | |
| 09:09 | I'm almost done with my latest toy: http://sol.arstecnica.it/ :-) | |
| 09:10 | it managed its first Carrom tourney without a glitch last saturday! | |
| 09:10 | kowey joined #darcs | |
| 09:10 | kowey | good morning! |
| 09:13 | gour | lelit: heh, cool. new job will be web-stuff? |
| 09:14 | twb | 19:39 <semka> is there any rss-feed or anything with recent "commits" to darcs.net repo? |
| 09:14 | gour | btw, darcs-fast-export claims it's capable for 2-way sync now |
| 09:14 | twb | semka: file a wishlist bug at bugs darcs.net |
| 09:14 | C-Keen | twb: there is the allmydata tracs but it is outdated |
| 09:15 | lelit | gour: interesting |
| 09:15 | twb | C-Keen: well, *I* don't care :-) |
| 09:15 | gnus' nnrss is crap | |
| 09:15 | C-Keen | true |
| 09:15 | gour | lelit: there is darcs-fast-export & import available, with convenient d2x x2d scripts |
| 09:17 | lelit | gour: let me say I have some doubt it works so simple, but I'm glad it's possible |
| 09:18 | that would also mean tailor's #1 ticket will never be closed :) | |
| 09:20 | gour | lelit: yeah, it looks simple... |
| 09:22 | * gour | --> out. bbl |
| 09:22 | semka | twb: ok |
| 09:23 | twb | Has anyone actually used darcs-f-e and d-f-i on large "zoo" repos? |
| 09:23 | e.g. GHC | |
| 09:36 | _keturn joined #darcs | |
| 09:42 | kowey joined #darcs | |
| 09:49 | nothingmuch joined #darcs | |
| 10:06 | gal_bolle joined #darcs | |
| 10:42 | malcolmw joined #darcs | |
| 11:34 | mornfall_ joined #darcs | |
| 11:41 | mornfall1 joined #darcs | |
| 12:11 | zooko left #darcs | |
| 12:11 | zooko joined #darcs | |
| 12:11 | zooko | hm, I should fix http://allmydata.org/trac/darcs-2/browser to get automatically updated... |
| 12:11 | lambdabot | Title: / – darcs |
| 12:12 | semka | zooko: reading my wishlist bugreport? |
| 12:12 | zooko | No, reading C-Keen's comments in this channel. |
| 12:12 | What's your wish? | |
| 12:14 | semka | rss feed from darcs.net repo (: |
| 12:14 | kowey | semka: would darcs weekly news on http://blog.darcs.net be enough? |
| 12:14 | lambdabot | Title: darcs advanced revision control system |
| 12:14 | kowey | oh wait, it doesn't actually list the patches :-) |
| 12:15 | semka | hehe |
| 12:17 | kowey | semka: maybe http://joyful.com/darcsweb/dar[…]-stable;a=summary |
| 12:17 | lambdabot | Title: joyful.com darcsweb - darcs-stable |
| 12:17 | kpreid_ joined #darcs | |
| 12:17 | kowey | cool, sm has darcs-graph integrated with that |
| 12:18 | semka | kowey: looks like it's what i need, thanks |
| 13:09 | kowey | anybody here willing to moderate darcs-users? |
| 13:09 | the job is basically to check the spam-trap for legitimate messages on a daily basis | |
| 13:24 | mib_dcmpfh joined #darcs | |
| 13:25 | mib_dcmpfh | darcs get http://code.haskell.org/~judah/ghci-haskeline/ results in Invalid repository: http://code.haskell.org/~judah/ghci-haskeline darcs failed: Failed to download URL http://code.haskell.org/~judah[…]/_darcs/inventory : HTTP error (404?) |
| 13:25 | lambdabot | Title: Index of /~judah/ghci-haskeline |
| 13:25 | mib_dcmpfh | I defined http_proxy variables. |
| 13:25 | Or doesn't darcs use them? | |
| 13:28 | kowey | if darcs is using libcurl, it should use them |
| 13:28 | (i'm also behind a proxy) | |
| 13:28 | darcs get of that repository works for me | |
| 13:29 | mib_dcmpfh: any idea how you built darcs? or where your binary came from? | |
| 13:29 | we use libcurl or libwww if they are available | |
| 13:59 | zooko left #darcs | |
| 14:19 | lisppaste2 joined #darcs | |
| 14:48 | semka joined #darcs | |
| 15:18 | kgoj joined #darcs | |
| 15:19 | kgoj | Hello! |
| 15:19 | I do darcs revert, and I see a big hunk | |
| 15:19 | I want to revert only part of it | |
| 15:19 | How can I achive this? | |
| 15:19 | how do I break hunk into smaller pieces? | |
| 15:20 | Igloo | You can't, I'm afraid |
| 15:20 | But if you manually revert one line at the boundary then darcs revert should show you 2 hunks | |
| 15:21 | kowey | i wonder if the hunk splitting interface that people are requesting for darcs record could be used here |
| 15:22 | kgoj | Igloo: thanks, I'll try it |
| 15:22 | kowey: Basically it's the same problem both when recording and reverting | |
| 15:23 | Igloo | kowey: Probably, but it doesn't exist :-) |
| 15:23 | kgoj | two changes I want to be separate get into one hunk |
| 15:26 | kowey | kgoj: then you'll be interested in http://bugs.darcs.net/issue126 then |
| 15:26 | lambdabot | Title: Issue 126: wish: ablity to split and otherwise record a hunk in record - Darcs b ... |
| 15:27 | * kowey | forgets why he hasn't applied the utf8 patch yet |
| 15:28 | kgoj | kowey: yes, it's exactly what I wished darcs had |
| 15:28 | Those 's' and 'e' from git commit -p | |
| 15:28 | kowey | great! well, we have C-Keen looking into it actually |
| 15:29 | C-Keen: still able to / interested in working on this problem? | |
| 15:40 | C-Keen | kowey: I need a better understanding of darcs' patches |
| 15:42 | kowey | would you like to re-assign yourself to other issues? |
| 15:42 | or are you feeling tenacious :-) ? | |
| 15:43 | (this could be a good way to acquire that understanding) | |
| 15:43 | gwern: you around? | |
| 15:44 | C-Keen | by feeling tenacious? |
| 15:45 | kowey | by working on this issue |
| 15:45 | I tend to find it easier to understand stuff if I have a concrete backdrop to frame them against | |
| 15:45 | C-Keen | I will keep at it, but please don't expect anything soon |
| 15:46 | maybe it might be easier if SelectChanges is refactored using a real zipper first, I don't know | |
| 15:46 | kowey | thanks! |
| 15:46 | work to make code cleaner before the work to actually do new stuff is always appreciated | |
| 15:47 | C-Keen | I have had trouble getting time for darcs |
| 15:47 | kowey | i'm not in a hurry for this feature, btw... I see this as the C-Keen training programme |
| 15:47 | C-Keen | :) |
| 16:05 | gbeshers joined #darcs | |
| 16:05 | Igloo | http://www.reddit.com/r/progra[…]ss_unique/c06nn3t - youtube got camp-dotted? |
| 16:05 | lambdabot | Title: citizenterminal comments on What makes the camp and darcs VCSs unique, http://tinyurl.com/5vynrw |
| 16:10 | C-Keen | Igloo: you might want to think of putting a coralized link of the video on the camp site |
| 16:11 | Igloo | A what? |
| 16:11 | C-Keen | coral is a content distribution network project where you prepend their server with your url |
| 16:11 | http://www.coralcdn.org/ | |
| 16:11 | lambdabot | Title: The Coral Content Distribution Network |
| 16:11 | C-Keen | so it gets cached the first time and then served via their network resources |
| 16:12 | Igloo | Surely youtube does somethign similar itself? |
| 16:13 | C-Keen | true, and the youtube video still works for me |
| 16:14 | Igloo | Yeah, I suspect it was a transient problem. Or perhaps it was really the projects.h.o page that wasn't loading |
| 16:16 | C-Keen | that one appears to be slow indeed |
| 16:30 | kowey | hmm, applying for immigration works like using software... it's really a case of rtfm sometimes |
| 16:30 | Igloo | Heh |
| 16:30 | kowey | i do wish the Border Agency would put something like an FAQ up though |
| 17:01 | zooko joined #darcs | |
| 17:02 | n6mod joined #darcs | |
| 17:02 | zooko | Dear people of #darcs: n6mod is the operations man for allmydata, and he has a darcs problem. |
| 17:02 | It has to do with reading gzipped patches, which I know was recently an issue. | |
| 17:02 | n6mod | I'm getting this: |
| 17:02 | amduserthumper:~$ darcs get zandrdev:/home/darcs/tahoe/server |
|
| 17:02 | Applying patch 1994 of 2853... Unapplicable patch: | |
| 17:02 | Mon Jan 28 11:56:22 PST 2008 warnerallmydata.com |
|
| 17:02 | * display the Helper FURL and our connection status on the welcome page. Closes #285. | |
| 17:02 | darcs failed: problem gzreading file /home/amduser/server/_darcs/patches/20080128195622-e01fd-a805b1db46503b277fe9dfdbb7978699240f65ec.gz | |
| 17:02 | zooko | darcs failed: problem gzreading file |
| 17:03 | What version of darcs are you using? | |
| 17:03 | n6mod | this is reproducible on multiple hosts. |
| 17:03 | amduserthumper:~$ darcs -v |
|
| 17:03 | 1.0.9 (release) | |
| 17:03 | zooko | There is a recent build of darcs which... I don't really know the details. |
| 17:03 | n6mod | installed from the hardy repository |
| 17:03 | zooko | Did kowey post a FAQ about this... |
| 17:03 | * zooko | looks. |
| 17:04 | n6mod | so, I have another copy of this tree that I believe is current in my home dir. let me see if I can 'get' that |
| 17:04 | nope, same error | |
| 17:05 | is this a backward-compat problem? | |
| 17:06 | Igloo | If 1.0.9 is giving the error then it doesn't sound like the recent problem |
| 17:06 | Does that file exist? | |
| 17:07 | n6mod | Yes, 429 bytes long |
| 17:07 | * Igloo | is a little confused; is that /home/amduser/server/... path in the repo that you are trying to create? |
| 17:07 | n6mod | yes |
| 17:07 | zooko | Ah, there it is |
| 17:07 | http://lists.osuosl.org/piperm[…]ember/016299.html | |
| 17:07 | lambdabot | Title: [darcs-users] workarounds for Codec.Compression.Zlib errors in darcs, http://tinyurl.com/69klal |
| 17:08 | Igloo | OK, what happens if you zcat it? |
| 17:08 | zooko | ^-- "[darcs-users] workarounds for Codec.Compression.Zlib errors in darcs" |
| 17:08 | n6mod | Looks like a patch, but... |
| 17:08 | gzip: server_1/_darcs/patches/20080128195622-e01fd-a805b1db46503b277fe9dfdbb7978699240f65ec.gz: invalid compressed data--crc error | |
| 17:08 | at the end of the output | |
| 17:10 | Igloo | Oh, maybe 1.0.9 also checks the gz checksums correctly |
| 17:10 | Has darcs 2 been involved at all? | |
| 17:10 | n6mod | yes, most of use darcs 2. |
| 17:10 | most of us | |
| 17:11 | Igloo | OK, so it probably is the recent problem then, and the recompressing dance should fix it |
| 17:11 | n6mod | I was hoping to be able to do a get using the darcs that comes in Hardy |
| 17:11 | Igloo | But it'll happen again if you don't upgrade the buggy darcs2 |
| 17:11 | n6mod | got a pointer to the dance steps? |
| 17:11 | zooko | Okay, so it sounds like the workarounds in that mail from Eric Kow are the way to go then. |
| 17:11 | n6mod: http://lists.osuosl.org/piperm[…]ember/016299.html | |
| 17:11 | lambdabot | Title: [darcs-users] workarounds for Codec.Compression.Zlib errors in darcs, http://tinyurl.com/69klal |
| 17:11 | zooko | Eric Kow is the new leader of the darcs project, by the way. |
| 17:12 | Igloo | You need to do something like "darcs optimize --uncompress; darcs optimize --compress" but I don't have a reference handy |
| 17:12 | * kowey | is mostly a bureaucrat in the great darcs adhocracy |
| 17:12 | zooko | Waitaminute |
| 17:13 | that link has nothing to help Zandr in it. | |
| 17:13 | It is about building darcs somehow-or-other. | |
| 17:13 | http://lists.osuosl.org/piperm[…]ember/016271.html | |
| 17:13 | lambdabot | Title: [darcs-users] workarounds for Codec.Compression.Zlib errors in darcs, http://tinyurl.com/6qeg2n |
| 17:13 | zooko | I accidentally clicked on the wrong message. |
| 17:14 | kowey | you're not building on Windows by any chance? |
| 17:14 | zooko | Hm, does *that* link have anything to help Zandr, in it? |
| 17:14 | We're not building at all. | |
| 17:14 | At least, we hope not. | |
| 17:14 | kowey | where are you getting darcs from? |
| 17:14 | building darcs *is* related to the problem | |
| 17:14 | zooko | Zandr wants to use the darcs-1.0.9 for linux/amd64 that came with the operating system (Ubuntu Hardy). |
| 17:14 | kowey | oh |
| 17:14 | then maybe it's not :-) | |
| 17:14 | zooko | Surely it is possible for him to do so, right? |
| 17:14 | n6mod | I think the problem was introduced by the darcs 2 running on Hanford. |
| 17:15 | 2.0.2 (+ 1 patch) | |
| 17:15 | kowey | so nsmod: are you getting Codec.Compression.Zlib errors at all? |
| 17:15 | zooko | n6mod: maybe try darcs optimize --uncompress |
| 17:16 | Igloo | zooko: You need to replace the broken darcs on the server with a working one, I believe |
| 17:16 | n6mod | so, trying to decompress on the repo in my home dir on hanford (our dev server) |
| 17:16 | Igloo | (where "the server" is where the repo you are trying to get is) |
| 17:16 | zooko | Um, what? The darcs-2 on hanford is bad? |
| 17:16 | Confusion reigns. | |
| 17:16 | n6mod | kowey: just the error gzreading when it tries to apply all the patches after a get |
| 17:17 | Igloo | Yes, it created a patch with a broken checksum, I believe |
| 17:17 | Although I have no experience of this problem, so someone please correct me if I'm wrong :-) | |
| 17:17 | mmorrow joined #darcs | |
| 17:17 | zooko | So, if anyone has pulled one of these broken patches, then what do they do? |
| 17:17 | They have to obliterate that patch and get a new one from me after I figure out how to produce fixed patches ? | |
| 17:18 | kowey | would this script help? http://lists.osuosl.org/piperm[…]09d/attachment.sh |
| 17:18 | lambdabot | http://tinyurl.com/5oyq2t |
| 17:18 | n6mod | so, I did darcs optimize --uncompress on the repo in my home dir. |
| 17:18 | Igloo | Recompress with a working darcs |
| 17:18 | zooko | What version of darcs should I run on our central server? |
| 17:18 | At the moment it is 2.0.2 (+ 1 patch) | |
| 17:18 | Igloo: is the darcs-1.0.9 on Zandr's machine "A working darcs"? | |
| 17:19 | What are the "non-working darcs's"? Is my 2.0.1+1? | |
| 17:19 | n6mod | I was then able to get from the darcs 1.0.9 machine successfully. |
| 17:19 | zooko | Yay! Problem solved. |
| 17:19 | kowey | zooko: this gzread problem is new to us actually |
| 17:19 | zooko | Oh, okay, I was wondering why you weren't answering my questions. :-) |
| 17:19 | I thought this was the same as the other problem. | |
| 17:19 | Any recommendations about what version of darcs to run on allmydata.com's central server? | |
| 17:19 | Igloo | I don't understand why n6mod could uncompress the patch with optimize --uncompress, but not when trying to apply it |
| 17:20 | kowey | I would recommend upgrading from 2.0.2 to 2.1.0 just as a general principle of the latter having fixed bugs in the former |
| 17:21 | but that's not related to your current issue | |
| 17:21 | hmm | |
| 17:21 | so it's darcs 1.0.9 that's having trouble reading patches that are produced by 2.0.2? | |
| 17:22 | n6mod_ joined #darcs | |
| 17:22 | zooko | Apparently. |
| 17:22 | n6mod_ | sigh... net issues here |
| 17:22 | zooko | And apparently darcs optimize --uncompress made the problem stop happening. |
| 17:22 | kowey | does darcs optimize --compress still allow darcs get to work? |
| 17:23 | zooko | n6mod_: http://codepad.org/21wqufwU |
| 17:23 | ^-- IRC log that you might have missed | |
| 17:23 | kowey | i think it *should* (this time with the patches being compressed in one shot instead of incrementally) |
| 17:24 | my recommendation is to upgrade to darcs 2.1.0 and yell at us if we continue to produce problematic patches | |
| 17:24 | (unless wiser people like Igloo have other suggestions) | |
| 17:25 | n6mod_ | OK, trying to optimize --compress the repo now, I'll try another get when that finishes |
| 17:25 | but I have a working tree, so I'm at least able to continue on my way. :) | |
| 17:26 | kowey | that's the most important thing, yes :-) |
| 17:26 | thanks for your patience with this! | |
| 17:26 | n6mod_ | Directory '/home/amduser/server' already exists, creating repository as '/home/amduser/server_3' |
| 17:26 | Applying patch 2 of 39... Unapplicable patch: | |
| 17:26 | Fri Jul 25 15:13:00 PDT 2008 warnerallmydata.com |
|
| 17:26 | * test_system.py: create SystemTestMixin, with less cruft, for faster system-like tests | |
| 17:26 | darcs: ./src/allmydata/test/test_system.py: openBinaryFile: does not exist (No such file or directory) | |
| 17:27 | So, two things bug me here. One is that it's a different error, the other is '2 of 39' | |
| 17:27 | As this repo had 2853 patches before the compress. :) | |
| 17:27 | kowey | so this is the result of optimize --compress? |
| 17:27 | n6mod_ | yes |
| 17:28 | I did an optimize --uncompress, and could then get a good tree | |
| 17:28 | optimize --compress (using the same 2.0.2+1) and then a get from the 1.0.9 machine results in the above error | |
| 17:29 | kowey | ok... are you willing to be extra-extra-helpful by trying optimize --compress with darcs 2.1.0? |
| 17:29 | n6mod_ | I could, sure. |
| 17:29 | kowey | (darcs 2.1.2 is fine; they are the same) |
| 17:30 | n6mod_ | zooko: what's the easiest/least disruptive way to get a darcs 2.1.x onto hanford? |
| 17:30 | zooko | Onto hanford? Why? It just so happens that I am doing that right now. |
| 17:30 | But it shouldn't .. | |
| 17:30 | * zooko | reads IRC log. |
| 17:31 | zooko | Ok, just a minute. |
| 17:31 | n6mod_: for your information darcs on hanford is managed under stow (see /usr/local/stow/) | |
| 17:32 | n6mod_ | Ah, cool |
| 17:32 | zooko | and I generally compile the darcs executable myself on hanford, which I just finished doing for darcs-2.1.2. |
| 17:32 | Okay, so now I'm about to do "cd /usr/local/stow && sudo stow -D darcs* && sudo stow darcs-2.1.2" | |
| 17:33 | n6mod_ | cool |
| 17:33 | zooko | Done. |
| 17:33 | So now /usr/local/bin/darcs is a symlink to /usr/local/stow/darcs-2.1.2/bin/darcs | |
| 17:33 | n6mod_ | right. darcs -v concurs. :) |
| 17:33 | zooko | "make test" passed |
| 17:34 | n6mod_ | so I guess I want to take the original repo from /home/darcs/tahoe/server, copy that somewhere convenient, and then do the --uncompress/-compress dance? |
| 17:35 | zooko | That sounds good. |
| 17:37 | n6mod_ | Same failure as above |
| 17:37 | (openBinaryFile: does not exist) | |
| 17:37 | zooko | Please run darcs check on that repo. |
| 17:38 | kowey | so to sum up, compressed patches produced with darcs 2 give you an error when you do a darcs get with 1.0.9? |
| 17:38 | (darcs 2.x) | |
| 17:38 | zooko | It isn't that simple. |
| 17:38 | semka joined #darcs | |
| 17:38 | zooko | We've been producing patches with darcs-2 and darcs getting them with darcs-1 for a long time. |
| 17:38 | n6mod_ | It appears that the --compress borks the repository |
| 17:39 | zooko | n6mod_: we don't particular care if it is compressed or uncompressed... |
| 17:39 | n6mod_ | darcs check produces: |
| 17:39 | darcs failed: fromSlurpFile: Couldn't find file "./src/allmydata/test/test_system.py" | |
| 17:39 | zooko | If it helps make things work better for you, just run "darcs optimize --uncompress" whenever you see this bug. :=/ |
| 17:39 | Okay, so darcs check failed. That's interesting. | |
| 17:39 | Please run "darcs repair" | |
| 17:39 | kowey | that's a good course of action |
| 17:39 | n6mod_ | same error |
| 17:40 | kowey | hmm |
| 17:40 | would you be willing to let one of us perform some sort of forensics on this repository? | |
| 17:40 | n6mod_ | So if I'm understanding what I'm seeing, darcs 2.0.2 created a patch that 1.0.9 can't gzread. |
| 17:40 | kowey | one thing I would like to see also is if darcs 2.1 can read these patches |
| 17:40 | n6mod_ | --uncompress using 2.0.2 works fine |
| 17:41 | --compress with either 2.0.2 or 2.1.2 destroys the repo | |
| 17:41 | zooko | The repo is public and open source. |
| 17:41 | kowey | oh great! |
| 17:41 | n6mod_ | Though that particular branch isn't exposed anywhere |
| 17:42 | this is the tahoe-server branch, which is only used internally | |
| 17:42 | zooko | It is already public. |
| 17:42 | Just nobody cares. | |
| 17:42 | http://allmydata.org/trac/tahoe-server | |
| 17:42 | lambdabot | Title: allmydata.org Tahoe-Server |
| 17:43 | kowey | ok, may I ask one of you to file a bug report with the findings from this IRC session? |
| 17:43 | n6mod_ | But... how does he get at the repo through darcs? i've been using ssh |
| 17:43 | zooko | PErsonally, I'm not sure what those findings are. |
| 17:43 | kowey | for now, I suppose we'll have to leave you with a working but uncompressed set of patches :-( |
| 17:43 | zooko | n6mod: the tahoe-server repo? On hanford? That actual part of the hanford filesystem is not public, you are right. ssh only. |
| 17:43 | There is a mirror of that repo automatically updated which is on .org, that I linked to there. | |
| 17:44 | I don't think patch compression makes much difference. | |
| 17:45 | kowey | hmm |
| 17:45 | i would be curious about the reading-with-darcs-2.1 experiment as well | |
| 17:46 | as I understand it, just doing darcs repair reveals problems already | |
| 17:46 | n6mod_ | OK, I have a tarball of the unmolested repo |
| 17:46 | that is, the initial state where 1.0.9 can't read it, before any mucking about with optimize | |
| 17:47 | kowey | very good |
| 17:47 | zooko | Upload it to tahoe. :-) |
| 17:47 | n6mod_ | if we were running tahoe on hanford, I would. :) |
| 17:47 | zooko | Seems like our "server" repo is corrupted. |
| 17:48 | You can use http://webapi.allmydata.com:8123 or http://testgrid.allmydata.org:3567 | |
| 17:48 | If you want. | |
| 17:48 | Here's an easy way to reproduce a problem: | |
| 17:48 | darcs get http://allmydata.org | |
| 17:48 | sorry | |
| 17:48 | lambdabot | Title: tahoe-lafs |
| 17:48 | zooko | hit enter too soon |
| 17:48 | darcs get http://allmydata.org/source/tahoe/server/ | |
| 17:48 | lambdabot | Title: Index of /source/tahoe/server |
| 17:49 | zooko | http://codepad.org/4bEuK7kq |
| 17:49 | n6mod_ | I will, but I need to fetch the file off hanford first... net sucks here, so it will be a few minutes |
| 17:49 | kowey | i can reproduce this with darcs 2.1.2 |
| 17:50 | n6mod_ | weird... that looks like the post-compressed error condition |
| 17:50 | zooko: did you do any darcs optimize steps on home/darcs/tahoe/server? | |
| 17:50 | zooko | n6mod: I guess the repo on allmydata.org is corrupted. |
| 17:50 | I have done in the past, yes. | |
| 17:51 | kowey | is this a partial repository of some sort? |
| 17:51 | your first patch is tagged tahoe-server-1.2.0, which leads me to surmise that it may be so | |
| 17:51 | in which case, it's not entirely surprising that darcs check fails | |
| 17:51 | zooko | Hm? The public http repo on allmydata.org you mean? |
| 17:52 | * zooko | looks. |
| 17:52 | Igloo | checking a partial repo should work, I think |
| 17:52 | zooko | Huh, it is. I didn't realize that. |
| 17:53 | Maybe I should go ahead and put a full copy there. | |
| 17:53 | n6mod_ | So the process that mirrors the repo over to .org seems to be producing the same behavior as the --compress |
| 17:53 | C-Keen | Igloo: how so? check does a replayRepo IIRC and that tries to get all the patches, doesn't it? |
| 17:53 | Igloo | It should start from the earliest checkpoint |
| 17:53 | kowey | right, but since checkpoints aren't propagated, darcs check of my local copy would fail |
| 17:53 | zooko | That process is just a "darcs push", run on .com, pointing at .org, over ssh. |
| 17:53 | Which is triggered whenever a patch is committed to the repo on .com. | |
| 17:53 | n6mod_ | weird |
| 17:54 | kowey | so the good news is that your non-gz-related error is now a bit less mysterious :-) |
| 17:54 | Igloo | kowey: Are you sure? Can't you do a darsc get --partial of your local copy? |
| 17:54 | zooko | Okay, so because the repo at http://allmydata.org/source/tahoe/server was a partial repo (by accident), that explains some of the problems? And I should just put a full repo there? |
| 17:54 | lambdabot | Title: Index of /source/tahoe/server |
| 17:54 | zooko | Doing so now. |
| 17:54 | kowey | I can't |
| 17:54 | C-Keen | --partial should not do anything with a darcs2 repo |
| 17:54 | n6mod_ | the error you get trying to get from the mirror on .org looks like the error I see doing a get from a --uncompress/--compressed version of the repo on hanford. |
| 17:55 | Igloo | It's not a darcs2 repo |
| 17:55 | kowey | and I remember people requesting that checkpoints be propagated |
| 17:55 | n6mod_ | it's not the gzread error |
| 17:55 | kowey | darcs 2 seems to have an annoying (or maybe a good) behaviour with partial repositories |
| 17:55 | oh no wait, scratch that comment | |
| 17:56 | Igloo | kowey: I can: http://hpaste.org/12690 |
| 17:56 | arjanb joined #darcs | |
| 17:56 | kowey | Igloo: oh wait, I forgot that I didn't do darcs get --partial in the first place |
| 17:57 | C-Keen | bye! |
| 17:57 | kowey | see ya |
| 17:58 | hmm, and now doing darcs get --partial gives : darcs failed: Bad checkpoint! | |
| 17:58 | Igloo: does darcs check work on your partial partial? | |
| 17:59 | Igloo | Yes |
| 18:01 | kgoj left #darcs | |
| 18:01 | kowey | huh |
| 18:02 | * kowey | could have sworn checkpoints are not propagated |
| 18:02 | Igloo | They aren't if you do a full get |
| 18:03 | kowey | ah |
| 18:03 | ah, right and checkpoints HAVE to be propagated for darcs get --partial to work | |
| 18:03 | ok, so if you darcs get --partial A P; then darcs get --complete P P2 | |
| 18:04 | darcs check in P2 should break | |
| 18:04 | Igloo | No, the get will fail |
| 18:04 | kowey | then darcs get P P2 implies --partial? |
| 18:05 | this is all sounding more robust than I remember --partial being | |
| 18:05 | Igloo | No, it fails (with 1.0.9 and 2.1.0 at least; just tested it) |
| 18:05 | kowey | huh |
| 18:06 | so then I wonder how weird situations like this creep up | |
| 18:06 | right, you /have/ to do darcs get --partial P P2 | |
| 18:07 | ok, then I retract all my claims about partial repositories and go back to just scratching my head | |
| 18:08 | but somehow we have the situation where darcs get of their tahoe repository works, sans checkpoint and only fetching 39 out of presumably a whole lot of patches | |
| 18:08 | with darcs thinking that this is a complete repository | |
| 18:10 | well, I have to run | |
| 18:10 | zooko, nsmod_: you have a copy of the complete repository, right? | |
| 18:10 | (i.e. with more than 39 patches) | |
| 18:11 | n6mod_ | I do. pushing it up to tahoe now, I'll have a link for you in a bit. |
| 18:12 | kowey | ok, well I won't be around for a few hours, but hopefully #darcs will be able to continue helping you |
| 18:12 | bye! | |
| 18:35 | zooko | Off I go to lunch. |
| 19:33 | Peaker joined #darcs | |
| 19:33 | Peaker | hey, why are darcs bundles not diff-formatted? |
| 19:59 | n6mod_ | So, here's the download link for our broken repo as a tarball |
| 19:59 | http://testgrid.allmydata.org:[…]3%3A10%3A99473758 | |
| 20:16 | markstos joined #darcs | |
| 20:17 | markstos | I'm considering using "prehook" with darcs record to run files through "perltidy" before recording them. How do I let the posthook now which files are about to be recorded? |
| 20:18 | mornfall | That sounds like something you don't want to do. |
| 20:18 | markstos | What do I really want to do then? |
| 20:18 | mornfall | Or, maybe I just misunderstood. |
| 20:18 | Will that modify the files, or just check them? | |
| 20:19 | markstos | I want to modify them, I think. .. possibly with a backup. |
| 20:19 | perltidy can do either. (just report or modify) | |
| 20:20 | mornfall | I'd probably stick with just report. |
| 20:20 | markstos | Ok, then how I convey the list of files I'm about record, so I can just report on them with my prehook script? |
| 20:21 | The documents for --prehook document no details of files available to it: | |
| 20:21 | http://search.cpan.org/~markst[…]l-QuickAdd-0.5_1/ | |
| 20:21 | lambdabot | Title: Mark Stosberg / ICal-QuickAdd-0.5_1 - search.cpan.org |
| 20:21 | markstos | Oh oops, wrong clipboard. |
| 20:21 | http://www.darcs.net/manual/no[…]10000000000000000 | |
| 20:21 | lambdabot | Title: Darcs commands, http://tinyurl.com/27c4xj |
| 20:22 | markstos | But if you like Google Calendar's "Quick Add" feature, but wished it work with standard ICS files, that Perl project might be of interest. :) |
| 20:22 | Maybe something like the posthook solution of exporting DARCS_FILES would work. ... but it looks like it isnt' implemented yet. | |
| 20:22 | mornfall | markstos: I don't think there's currently a way, you might need to modify src/Darcs/Test.lhs. |
| 20:23 | Or, wait. | |
| 20:23 | * markstos | waits |
| 20:23 | * mornfall | looks around for the envvar exports. |
| 20:26 | mornfall | Hm, tough luck. |
| 20:26 | It seems that record currently does not export anything. | |
| 20:26 | push, pull and so on has DARCS_FILES env set in posthook... | |
| 20:27 | markstos | At least the code is consistent with the docs here. |
| 20:27 | mornfall | Moreover, it seems that these aren't available in the prehook anyway. |
| 20:27 | One way around that is a wrapper around darcs binary, and some programs go that route. | |
| 20:27 | Although it's quite inelegant... | |
| 20:27 | markstos | Then... I wonder how people are actually using prehook then. Like, isn't something using a "tidy" program of some sort, or doing line-ending conversion? |
| 20:28 | mornfall | I think these are generaly dangerous uses of prehook. |
| 20:28 | markstos | Maybe I'm approaching this wrong. I found no example of in-use prehook scripts by googling, though. |
| 20:28 | mornfall | You would need a better mechanism to implement that, and it'd still likely be fragile. |
| 20:29 | davidL joined #darcs | |
| 20:29 | markstos | OK |
| 20:33 | nomeata joined #darcs | |
| 20:35 | gour joined #darcs | |
| 20:47 | mornfall | markstos: I'd consider prehook a viable mechanism to enforce policies, but modifying files is icky. |
| 20:48 | markstos: First, you could get easily swamped with spurious changes in interactive record, second two people with slightly different versions of the program in question might end up flip-flopping some differences. | |
| 20:48 | zooko | It might be safer to have a .. oh, what mornfall said. To check if the file satisfies a policy and reject the patch if not. |
| 20:49 | markstos | mornfall: The .perltidyrc file would ship with the project and be referenced by the prehook, so we'd all be using the same policies. |
| 20:50 | But even if I just want to enforce policies... there's no way to get the list of files to enforce the policies on. | |
| 20:51 | mornfall | markstos: Well, you could run it in apply somewhere, and people would easily get chance to amend their patches and re-send/re-push. |
| 20:51 | Or so I understand the idea. | |
| 20:53 | markstos | mornfall, but doesn't prehook in darcs apply have the same limitation...no access to the file list? |
| 20:54 | mornfall | Dunno. |
| 20:54 | Well, true. | |
| 20:54 | But you can run it on everything, right? | |
| 20:55 | markstos | First, I'd rather not run it on over 300 files when only one or two changes. Second, I only want to run it on Perl scripts, not on the hundreds of HTML and image files, which would be useless. |
| 20:56 | mornfall | markstos: Well, the latter is a non-issue, just check the file for being perl. The former, well, yeah. |
| 20:56 | But it might as well be premature optimisation. | |
| 20:56 | Dunno. | |
| 20:56 | I probably don't have very good answers... | |
| 20:56 | markstos | Ok, I figured this out. |
| 20:57 | mornfall | It wouldn't be extremely hard to implement in darcs, but noone has done that so far. |
| 20:57 | markstos | the prehook can call "darcs whatsnew -s" to get the list of files. |
| 20:57 | An example using git does the equivalent: | |
| 20:57 | http://use.perl.org/~Yanick/journal/35626 | |
| 20:57 | lambdabot | Title: Journal of Yanick (3196) |
| 20:58 | mornfall | Well, it is a little icky, but you could do just that, indeed. |
| 20:58 | It specifically fails to take into account the possibility that you are only recording a subset of the files. | |
| 20:58 | markstos | It's less icky than processing 300 files to check one... |
| 20:59 | True, but it seems that it's the best that's possible with darcs. | |
| 20:59 | mornfall | Well, in record, yes, checking 300 files is more or less out of question. On a server where people push, it might be no problem. |
| 21:00 | markstos | We generally record and push all on the same system. |
| 21:00 | mornfall | Also, you can do a little trick, where you implement a record posthook, which looks at darcs changes --last 1 -s and examines those files, warning you that you should amend fixing this and that issue. |
| 21:01 | Possibly their pristine copies, so you don't compromise interactive records. | |
| 21:01 | (So you can interactively record good parts of bad files.) | |
| 21:02 | markstos | That could work. |
| 21:03 | mornfall | (At least I would find that the least intrusive in my normal workflow, although it might be hard to enforce that people do not ignore such warnings). |
| 21:03 | If you manage to get that work, it might be worth documenting somehow, too, btw. | |
| 21:06 | mjrosenb | copying a directory that's under version control is the same as darcs put, right? |
| 21:07 | markstos | It would be nice if "whatnew" had a mode to supply a plain list of file names for cases like this, without the leading "M", "A", etc, and without the plus/minus line changes. Currently when I do "darcs w -qs" and "darcs w -s"... the output is the same. Perhaps the "--quiet" mode could be updated to work this way, or maybe "--plain" or "-only-filenames" are even better. |
| 21:07 | mjrosenb: Not exactly, but it's absolutely safe and reasonable. I do it regularly. | |
| 21:07 | If you "put" to the same file system, you could safe space with hardlinks. | |
| 21:08 | Just being able to copy or move darcs repos so easily is one of the beautiful features it has. | |
| 21:08 | mjrosenb | hardlinks are automatically unlinked on write by the OS? |
| 21:09 | Peaker | I'd assume the hard links are for the darcs-internal files, and not the actual working tree? |
| 21:09 | (and then darcs can unlink them first) | |
| 21:09 | (but that's just an assumption) | |
| 21:09 | mjrosenb | that makes more sense |
| 21:10 | although having hardlinks would probably make this easier on my | |
| 21:10 | Peaker | we need revision control supported in the file system itself, IMO (and we need to replace the file system with typed objects, and generally rewrite software :-) |
| 21:11 | mjrosenb | diff_patch :: a -> a -> a -> a |
| 21:12 | or if you want a bit more abstraction | |
| 21:12 | diff :: a -> a -> Patch | |
| 21:12 | patch :: Patch -> a -> a | |
| 21:14 | markstos | My experience with the hardlinking policies is that they have tended to "just work" for me. |
| 21:14 | I don't think there's much risk involved. | |
| 21:15 | Peaker | mjrosenb: that looks deceivingly symmetrical |
| 21:15 | mornfall | I don't think put does hardlinks. Get does. |
| 21:15 | Peaker | Patch -> (a -> a) vs a -> (a -> Patch) |
| 21:15 | mornfall | I still think put is a misfeature. |
| 21:16 | Peaker | was darcs the first dvcs? |
| 21:16 | mjrosenb | Peaker: either that or mercurial. don't know which was first |
| 21:16 | pretty sure git was later as well | |
| 21:17 | Peaker | git was based on BitKeeper, which was also a kind of dvcs, I think |
| 21:17 | s/based/inspired | |
| 21:17 | arjanb | git is based on monotone |
| 21:18 | Peaker | monotone was horrid |
| 21:18 | markstos | I remember "arch / tla" being one of the early ones. |
| 21:18 | mjrosenb | hrmm, i guess patch should be patch :: Patch -> a -> Maybe a |
| 21:19 | : ( | |
| 21:19 | goodbye symmetry | |
| 21:19 | markstos | There was a time when darcs was being considered for the Linux kernel...but it doesn't scale up for projects that big, it turned out. |
| 21:19 | Peaker | mjrosenb: probably Either for more descriptive errors |
| 21:20 | markstos | ".bak" had a distributed mode, where I could send friends a file, and they could also make ".bak" copies, but like the centeralized ".bak" system, it had significant weaknesses. |
| 22:05 | markstos left #darcs | |
| 22:47 | conal joined #darcs | |
| 22:47 | Peaker | hey, how do I see/know what recorded patches I have on a local branch that are not in my pull location? |
| 22:48 | also, can I get darcs send to just dump a bundle as text, instead of actually mailing it? | |
| 22:59 | wjt | -o filename |
| 22:59 | or -O to dump to a file in the current dir | |
| 23:01 | Peaker | ah, its so interactive already that it could be just letting you type nothing in the email question for -O |
| 23:35 | twb joined #darcs |
← Previous day | Index | Channel Index | Today | Next day → | Search | Google Search | Plain-Text | plain, newest first