Camelia, the Perl 6 bug

IRC log for #darcs, 2008-04-18

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

All times shown according to UTC.

Time Nick Message
00:14 lelit what's the meaning of 's' and 'f' in darcs2 push? the help says s: don't push the rest of the changes to this file
00:14 lelit f: push the rest of the changes to this file
00:14 lelit
00:14 lelit and, btw, why English.lhs fails to spell the plural of "patch" as "patchs", in that same help?
00:23 sm lelit: s for skip, f for file
00:25 sm not sure about English.lhs.. that module is blowing my mind a little
00:26 sm "patches" would be correct
00:27 sm micro update: I added how to edit help at bottom of the wiki page
00:27 sm then I tried to do the same to the web page, which is darcs/doc/index.html.in
00:28 sm first I'm trying to indent and make that page valid x/html .. but I don't seem to have the *ml-fu
00:30 sm has another go.. finish it, finish it!
00:31 lelit sm: I still don't understand... I'm pushing whole changesets, what does it mean "don't push the rest of the changes to this file"? which "this" file??
00:32 sm lelit: I see, good question. s and f don't seem applicable to push
00:32 sm what does ? say
00:33 lelit I pasted that: "the help says..."
00:34 lelit and about English.lhs, yes, evenif it seems simple, Haskell is very hard when you miss little details...
00:34 lelit at first, the construct
00:34 lelit  plural (Noun s) | last s == 'e' = showString s .  showChar 's'
00:34 lelit  plural (Noun s) = showString s . showString "es"
00:34 lelit
00:35 lelit confuses me
00:41 sm lelit: figure it out ?
00:43 sm I figured it's a combination of pattern match and guard. plural s  is s followed by 's' if s ends with e. otherwise plural s is s followed by 'es'
00:57 lelit yes
00:57 lelit and what does produce "patchs"? :-)
01:02 lelit zookofamilytime: I'm pushing a set of patches on trac-darcs, with support for darcs-2 "query contents"; this hopefully should be quicker than the current hacky way
01:25 sm yay, a darcs patch!
01:25 sm it has been a while
01:26 sm in fact I wonder how I'd find my patches in darcs history
01:26 sm matches or patches, I believe
01:27 sm I find these distressingly hard to master
01:27 lelit darcs query authors | grep -i simon ?
01:27 lelit :)
01:27 sm right.. now to see the individual changes I figure darcs changes --matches or --patches ...
01:27 lelit --match "author foo"
01:29 lelit darcs is great, and darcs2 even greater!-)
01:29 lelit ok, so sleepy now...
01:29 lelit g'night sm
01:29 sm thanks lelit! sleep well
01:29 lelit thank you
01:30 sm darcs changes --matches 'author simon@joyful.com' .. lovely
01:30 lelit yep
01:31 sm I suppose there's no way to repair patches which have just the bare email address, at this point
01:31 lelit no
01:31 lelit the author is used to compute the hash of the patch
01:32 lelit darcs has a "complicated" map to build the changelog, grouping the various addresses of a single person
01:33 lelit at build time, to create the changelog
01:33 lelit ciao!
01:47 sm hmm.. how long does it take for a darcs send to show up on the dev list ?
01:47 sm the more immediate that is, the better
01:48 sm 20 minutes and waiting >:/
01:56 sm has fun with patch selection
01:56 sm patches recorded each year since 2002: 257, 3376, 4866, 4520, 1747, 2798, 2416
01:58 sm does darcs record the date a patch was added to a particular repo ? I am guessing not
01:59 sm so 2004 was a peak, 2006 took a dive, and 2008 is looking rather good!
02:00 clanehin sm: how did you determine that?
02:01 sm for y in 2 3 4 5 6 7 8; do echo "200$y" `darcs changes --matches "date 200$y" | wc -l`; done
02:03 sm on the darcs 2 repo, that takes 12s on my macbook
02:03 clanehin Run darcs changes --matches "date 2008" and manually examine the number of patches, please.   :)
02:04 sm oh shoot :)
02:04 sm well just divide those numbers by 4 :)
02:04 sm by 3
02:05 sm I believe this is better:
02:05 sm 2002 86
02:05 sm 2003 1010
02:05 sm 2004 1378
02:05 sm 2005 1225
02:05 sm 2006 373
02:05 sm 2007 800
02:05 sm 2008 657
02:05 clanehin I don't see anything obvious to grep again, since patch record data seems to appear in some patch descriptions.
02:06 clanehin But yes, this is good news for darcs.
02:06 sm hmm.. I tried  egrep '^[^ ]' | wc -l
02:06 sm ah! darcs changes --count, of course!
02:07 sm yes, those are accurate
02:09 sm darcs process improvement idea #43: keep a darcs repo activity graph on the front page
02:09 clanehin sm: he kindof did that, by indicating the number of patches in the version string in the title of the manual
02:10 clanehin http://darcs.net/manual/
02:10 sm that's just indicating the version, no ?
02:10 sm it's relative to the release
02:10 clanehin Yes, but it advertises activity
02:11 sm and it's precise, which is nice
02:18 sm oh frickin 'ell.. my provider is blocked by sorbs
02:19 sm more work required to let me darcs send
02:19 sm well, good evening all
04:01 lispy good evening
04:03 lispy sm: I like the graph on the front page idea
04:09 zookofamilytime too.
04:10 zooko So I was wrong to think that I could run darcs-2 executable, and hashed-format repos on a server and let darcs-1 executables interoperate with it.
04:11 zooko See all these red boxes? http://allmydata.org/buildbot​/waterfall?show_events=false
04:11 zooko Hm.. wait a minute.
04:11 zooko I put the darcs-1 repos back and it is *still* red...
04:16 lispy Hmm
04:16 lispy zooko: it turns out, I don't know how to read that
04:17 lispy What do you think happened between darcs1 and darcs2
04:17 lispy ?
04:18 zooko Now, I didn't.
04:19 zooko Sorry.
04:19 zooko What I meant was:
04:19 zooko No, I didn't put the darcs-1 repos back yet, when it was still red.
04:19 zooko lispy: I thought that the darcs-1 executable talked to the darcs-2 executable and asked it for patches, and the darcs-2 executable gave it back something it could use.
04:19 zooko Silly me.
04:20 lispy Er, I don't know how to read the waterfall thingy...(I see what I said was really ambiguous)
04:22 zooko Vertical axis is time (most recent at the top), columns are build jobs.
04:23 zooko Red means failure.  Green means happiness.
04:24 lispy It kinda looks like "cygwin" is a time
04:24 lispy But, cygwin is a buildjob?
04:27 zooko Why does it look like a time?  Yes, the "cygwin" column is the history of one builder.
04:28 lispy Well, to my untrained eyes, I wanted to read it like a matrix, (or like Excel) and I thought that "time" was telling you the title of the values you'd see in that row
04:28 zooko Oh, I see.
04:28 zooko Thanks for the feedback.
04:34 zooko Hm.  Can you think of a way to express the information representation design issue that tripped you up?
04:35 zooko I'm thinking that if the label "time" applied to the column rather than the row (as it does),then there ought to be an underline under it rather than a bar to the right of it.
04:35 zooko That's the way spreadsheets do it, right?
04:35 zooko I started to submit a ticket on buildbot.net, but I'm unsure of how to explain the relevant UI principle.
04:40 lispy zooko: I'd probably fix it by making the time column special
04:40 lispy zooko: it is unique, so let's just make it visually unique
04:40 lispy Like one solid color or something
04:41 zooko That sounds good.
07:55 lelit g'morning chx!
08:00 chx lelit: hey
08:02 lelit hi
08:12 chx lelit: any idea how long the resync will take?
08:13 lelit lemme see
08:13 chx An excerpt from ps aux:
08:13 chx R    00:37 155:17 /usr/local/bin/python /usr/bin/trac-admin /home/trac/nd resync
08:16 lelit it took an hour here
08:16 chx can it be hung or somethin?
08:17 lelit did you look in the log?
08:18 lelit there's now an handy "about to load patch 2/2345"
08:21 chx log... log...
08:22 lelit did you enable it?
08:22 chx poor thing
08:22 chx it stil runs
08:23 chx the log is 11M so far...
08:23 lelit mine is twice that
08:23 lelit :)
08:24 lelit but I did several runs yesterday
10:46 lelit chx, how is going?
10:47 chx lelit: steadily!  R    00:37 308:45 /usr/local/bin/python /usr/bin/trac-admin /home/trac/nd resync
10:47 chx lelit: this is no speed daemon.
10:48 chx i now recognize the patches it loads
10:48 chx it's nearing today :P
10:50 wferi lelit: Hi! Why don't you accept my TracDarcs docu suggestions about disabling SVN and giving a hint for the global config?
10:54 lelit wferi: time?
10:54 wferi possibly, I just don't know.
10:54 lelit I spent all day yesterday on it...
10:54 lelit did you try the latest support for darcs 2?
10:55 wferi On developing TracDarcs?
10:55 lelit yes
10:55 wferi No, I didn't.
10:55 wferi Well, thank you very much.
10:55 wferi I really did mean to be mean.
10:55 wferi s/did/did not/
10:56 lelit sorry, but what does that mean? I mean, "mean to be mean"? sorry for the redundancy :)
10:56 lelit takes the dictionary
10:57 wferi all-dict mean
10:57 wferi lambdabot does have sth like that...
10:58 wferi Anyway, it means that I very much appreciate your work and don't want to annoy you.
10:58 lelit oh, I see :)
10:58 wferi Just trying to live to my promise of documenting my experiences.
10:59 lelit learnt a new "meaning" of mean :)
10:59 wferi But creating a separate page doesn't seem like the best solution...
10:59 wferi Including that into yours is, in my opinion.
10:59 wferi :)
11:01 wferi what happened to lambdabot?
11:05 chx lelit: you would not believe this . finished.
11:11 chx lelit: but /browser does not want to load for me. http://darcs.pastebin.com/d2aa776ad
11:23 lelit maybe I screwed up something
11:23 lelit does the log say anything else?
11:24 chx not really
11:24 chx DEBUG: cd /home/darcs-master; TZ=UTC darcs diff --diff-command "cat %2" --to-match "hash 20070308172038-5c0
11:24 chx 39-3111c4243b737f9c46dd5f71355efb2f59f6a2e5.gz" "tailor.config"
11:24 chx this was the command before it.
11:27 lelit chx, then, weren't you using darcs-2?
11:27 chx lelit: i am
11:27 lelit chx, it seems a problem with an SQL
11:28 lelit uhm, very strange
11:28 lelit in the log, when the trac starts, you should see something like "Using darcs, version X.Y.Z"
11:28 lelit is it there?
11:30 chx looking
11:30 chx no such thing
11:30 chx i have log set to DEBUG
11:31 chx I am using 0.10.4 trac
11:33 chx with that patch unpulled suggested 2 days ago
11:45 lelit everything seems right here
12:44 chx lelit: LOVE YOU
12:44 chx lelit: downloaded sqlite3 , configured , installed, reconfigured and reinstalled python.
12:45 chx lelit: _WORKS_!!!!
13:47 nornagon so i'm trying to write an apply posthook that will get an IRC bot to spit out commit messages
13:47 nornagon the darcs docs say things about a DARCS_PATCHES_XML environment variable being set
13:47 nornagon but it never is
14:04 lelit chx: you're welcome!
14:04 chx lelit: it's not the world's fastest to say the least...
14:04 chx lelit: but it will do
14:05 chx lelit: i will work on getting a much beefier server
14:07 lelit what kind of machine is that?
14:26 lelit zookosleep: doesn't darcsver have an homepage?
14:27 lelit http://pypi.python.org/pypi/darcsver points to a non-existing one
14:27 lambdabot Title: Python Package Index : darcsver 1.1.3
14:28 lelit I get an error calling the standalone script, "darcsver"
14:43 lelit fyi, I tagged version 0.6 of the tracdarcs plugin
14:56 lelit hi chx, about the performance, I bet trac 0.11 is something better (exactly for that patch that you unpulled): 0.10 checks for new patches more often than strictly needed
14:56 lelit and in your case, "darcs changes" is a big hit
14:57 lelit maybe we should find a lighter way to check for new patches
15:01 chx maybe :)
15:01 chx what can I do server wise to make it a smaller hit?
15:01 chx put the darcs repo on a ramdrive?
15:02 chx buy a (much) faster CPU ? would lots of cores help?
15:02 chx in short, what bounds
15:02 lelit I think it's mostly CPU bound
15:03 lelit on my amd64 it's usable
15:03 lelit ciao
15:04 chx bye
15:11 sm morning
16:01 zooko Morning!
16:02 zooko lelit: darcsver hasn't a home page.  I intend to add it into the darcs-tools project, I guess.
16:02 zooko nornagon: I use cia.vc for that.  Visit #tahoe to see it in action.
16:03 zooko lelit: Sorry to hear that the standalone script is broken.  I don't use it since I use the setuptools plugin everywhere now.
16:03 zooko Could you paste the error message, possibly to http://tools.darcs.net/trac/darcs-tools ?
18:40 zooko Hm.  I'm going to re-arrange the list of prebuilt binaries for Mac: -SOURCE = "buildslave@dev.allmydata.com:/home/darcs/tahoe"
18:40 zooko -SOURCE_HTTP = "http://allmydata.org/source/tahoe/trunk"
18:40 lambdabot Title: Index of /source/tahoe/trunk
18:40 zooko oops, wrong paste.
18:40 zooko prebuilt binaries for Mac: http://wiki.darcs.net/DarcsWiki/CategoryBinarie​s#head-c14449f675f1f36c70703538d40e79f056a69bd9
18:40 lambdabot Title: CategoryBinaries - DarcsWiki, http://tinyurl.com/3ycs9p
18:48 zooko There.  Please feel free to review my changes: http://wiki.darcs.net/DarcsWiki/CategoryBinarie​s#head-c14449f675f1f36c70703538d40e79f056a69bd9
18:48 lambdabot Title: CategoryBinaries - DarcsWiki, http://tinyurl.com/3ycs9p
18:52 lelit hi zooko
18:55 zooko Hi lelit!
18:55 lelit already tried latest trac-darcs? :)
18:56 zooko Not yet.
18:56 zooko I'm looking forward to it, though.
18:56 zooko I'm working on darcs, buildbot, etc for allmydata.org.
18:56 zooko darcs-2 can be so much faster than darcs-1, it's great.
18:56 zooko http://allmydata.org/buildbot​/waterfall?show_events=false
18:56 lelit yeah
18:56 zooko ^-- look at the "nooxie" column.
18:57 zooko if you click on the darcs checkout rectangles, the log includes "elapsedTime" which tells how many seconds each unix command takes...
18:57 zooko Sorry, I mean the "solaris" column, not the "nooxie" column.
19:00 lelit zooko: is there at least a repo for darcsver?
19:03 zooko lelit: yes
19:04 zooko looks for where it is available...
19:04 zooko makes it availabler...
19:05 lelit are you going to host various projects on that trac?
19:05 lelit if so, with a single repo, or some hack to point to multiple ones?
19:06 zooko The darcs-tools trac?
19:06 lelit yes
19:06 zooko My current thought was to throw all the orphaned or small darcs tools into one repo.
19:06 zooko What do you think?
19:07 zooko http://allmydata.org/source/darcsver/trunk/
19:07 lambdabot Title: Index of /source/darcsver/trunk
19:07 zooko ^-- darcsver repository
19:07 lelit yes, that's my preference too, for such small pieces
19:08 lelit it would be nice and wise to *mandate* that each patch doesn't touch multiple "projects"/"subdirs of the root dir"
19:08 lelit so that, whenever the need arise, we can get a clean and standalone darcs repo
19:08 lelit for any single project I mean
19:09 zooko That makes sense.
19:10 lelit once upon a time, well before I gave up with the idea for my own projects, I even scratched a check script, but it's surely gone
19:11 zooko nods.
19:12 lelit when I started with tailor (actually, with cvsync) I tried to merge in a single darcs repo the various "products" that compose Plone (for example), but at that time it didn't scale very well (read, darcs 0.x took ages..)
19:13 zooko :-)
19:14 lelit of course, the ideal path would be adapting trac-darcs to the multirepos-branch of trac... ;)
19:14 sm a darcs 2 mirror of zope & plone stuff would be interesting
19:15 lelit sm: http://darcs.arstecnica.it/3rd/
19:15 lelit ;-)
19:15 lambdabot Title: Third party's stuff
19:15 sm that reminds me, darcs needs to grow really nice integration with svn, git etc.
19:16 lelit beware, that's not up-to-date, but I do keep all my 3rd party pieces there :)
19:16 sm like bzr and git do.. a big selling point. Do you think a chunk of tailor could be moved into darcs for that ?
19:16 sm woah, nice lelit
19:17 zooko lelit: that would be very interesting!
19:17 zooko multi-repos/multi-branch trac-darcs!  :-)
19:18 lelit sm, I never used those "mixin", and I don't know how useful they are... but I think there's nothing in tailor that could be implanted in darcs...
19:18 lelit anyway, dinner time!
19:18 lelit later, ciao
19:18 sm well, for example darcs convert svn://......
19:19 sm darcs pull svn://...
19:19 sm etc.
19:19 lelit zooko: http://trac.edgewall.org/b​rowser/sandbox/multirepos
19:19 lambdabot Title: /sandbox/multirepos – The Trac Project – Trac
19:19 lelit sm, I see the use case, I do not see how tailor code could be reused, if not darcs executing an external tool...
19:20 sm is tailor python ?
19:20 lelit dunno if there are python bindings for haskell ;-)
19:20 sm there are
19:20 sm if not directly, perhaps at least your designs could be used
19:20 lelit in a previous life I started PyObjC and it was fun (now standard part of Cocoa)
19:21 sm nice!
19:21 lelit :)
19:21 sm ikiwiki, though in perl, allows plugins in any language via a xml-rpc interface
19:22 lelit but I keep missing a lot of little pieces reading/understanding haskell, so surely it will be in my *following* life, if any :)
19:22 sm an interface for darcs plugins might be stimulating
19:22 zooko :-)
19:22 lelit disappears
19:23 zooko ciao!

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