Camelia, the Perl 6 bug

IRC log for #darcs, 2010-01-27

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

All times shown according to UTC.

Time Nick Message
00:05 twb They're probably using doxygen
00:05 twb The end goal is for our documentation to end up looking like this: http://sphinx.pocoo.org/contents.html
00:06 twb That implements searching by having js download what is basically a TAGS file built from the html docs
00:06 twb So you get searching using a static http server, which is cute (if annoying for me personally).
00:07 lispy That is a nice looking website.  Allowing user comments would be icing on that cake.
00:07 twb These documents (but probably not the front page) are also built using rst: http://docs.python.org/3.1/
00:08 twb I also worked out how to auto-generate such a static set of HTML from our existing gitit wiki
00:09 twb http://pastebin.com/f63a80e10 are the relevant files from my test instance.
00:12 twb mornfall: ping.
00:19 mornfall yes?
00:19 mornfall (But I'm about to sleep.)
00:22 twb mornfall: sorry, I put it in email
00:22 twb mornfall: summary: siva doesn't like cabal 1.8
01:10 kpreid_ joined #darcs
01:26 gbeshers joined #darcs
02:30 kpreid_ joined #darcs
03:52 Igloo_ joined #darcs
06:16 JaffaCake1 joined #darcs
07:08 JaffaCake joined #darcs
07:08 merehap left #darcs
07:22 balor joined #darcs
07:56 sea-gull joined #darcs
08:02 darcscommitbot joined #darcs
08:03 darcswikibot joined #darcs
08:18 lelit joined #darcs
08:54 Gowilla joined #darcs
08:54 lelit joined #darcs
08:54 darcswikibot joined #darcs
08:54 darcscommitbot joined #darcs
08:54 sea-gull joined #darcs
08:54 JaffaCake joined #darcs
08:54 Igloo_ joined #darcs
08:54 kpreid_ joined #darcs
08:54 gbeshers joined #darcs
08:54 sshc joined #darcs
08:54 Korusef joined #darcs
08:54 kolibrie joined #darcs
08:54 sm joined #darcs
08:54 thorkilnaur joined #darcs
08:54 kosmikus joined #darcs
08:54 dons joined #darcs
08:54 tmug joined #darcs
08:54 felipe joined #darcs
08:54 Jaak joined #darcs
08:54 C-Keen joined #darcs
08:54 gwern joined #darcs
08:54 PHP-Pierre_ joined #darcs
08:54 karlpinc joined #darcs
08:54 kolmodin joined #darcs
08:54 dleverton joined #darcs
08:54 |Mike| joined #darcs
08:54 mornfall joined #darcs
08:54 npouillard joined #darcs
08:54 idnar joined #darcs
08:54 lambdabot joined #darcs
08:54 dcoutts joined #darcs
08:54 nwf joined #darcs
08:54 gh_ joined #darcs
09:03 gal_bolle joined #darcs
09:04 npouillard joined #darcs
09:08 npouillard joined #darcs
09:10 npouillard joined #darcs
09:36 kowey joined #darcs
10:23 kowey are we linking to darcs.net/releases from anywhere?
10:28 kowey perhaps I should figure out how to enable directory listing on that
10:28 malcolmw joined #darcs
11:30 kowey hmm, beta2 doesn't seem to do much better than beta1 on the check/repair stuff (from lispy's) benchmarks
11:31 kowey (which is OK, imho)
11:32 mornfall I wish lispy could format his emails a little more readably. The wrapped tables are useless as far as I am concerned.
11:32 kowey yeah
11:33 * kowey suspects he's copy+pasting into gmail
11:33 kowey the zip file helps, though
11:33 kowey @tell lispy btw, my favourite trick on macosx is to pbcopy < file and then paste from the Mac clipboard
11:33 lambdabot Consider it noted.
11:33 mornfall Ah, ok.
11:34 kowey on my agenda when I have time is to write an RST-style renderer for darcs-benchmark (no need for it to be in tabular)
11:34 kowey and also to tweak --dump to output JSON so that Max et al can do fun things with Python
11:34 mornfall Eh.
11:35 kowey not a good use of effort?
11:35 kowey I guess the JSON effort is slightly wasted because we'd ideally just be able to do everything from darcs-benchmark
11:36 mornfall Well, it will just make it hard to process the data in Haskell.
11:36 mornfall If we store it in JSON.
11:36 kowey and text-formatted stuff is simplest because it allows for simple copy+pasting
11:36 kowey oh, well you just implement a class to read
11:36 kowey what makes it not so painful is there there are already lots of instances for basic data
11:37 mornfall As opposed to deriving?
11:37 kowey hmm?
11:37 kowey oh, you're saying that Read just makes life easier because it's auto-derived?
11:37 mornfall I don't really see the benefit, unless someone has concrete plans to implement a presentation of the benchmarks in non-Haskell.
11:38 kowey yeah
11:38 kowey and also the text-is-simplest argument makes that less attractive
11:38 kowey I think my goal is to make it easy for people to copy and paste output into the wiki
11:38 kowey so it would have to be indented 4 spaces to look like verbatim text for now
11:38 mornfall Which is not very useful, since we want the presentation to be automated, right?
11:39 kowey well, long term objective is that you want fancy presentation with HTML and bar charts and what not
11:39 mornfall I.e. dump the data into a directory and have something render it properly.
11:39 kowey which is why I figured JSON would be good because we can then outsource that
11:39 kowey but on the short term, I thought a dump textual table would be the fastest route to having things that we can share/look at
11:39 mornfall As long as you believe we *will* outsource it, sure.
11:40 mornfall I may be too sceptical. :)
11:40 * kowey can only cross his fingers
11:40 kowey OK, I'll focus on the ad-hoc renderer for darcs-benchmark so that we are up and running asap on the short term
11:40 kowey and worry about the JSON later
11:40 mornfall You can have a JSON renderer if you like, but I'd prefer the --dump output to be readable by darcs-benchmark.
11:41 kowey oh yeah, that's a 3rd option
11:41 kowey I just meant to implement something that's like Text.Tabular.AsciiArt, but with some minor twists
11:41 kowey OK, well, I'll bang something out and show you what I mean
11:41 kowey err... later...
12:13 malcolmw joined #darcs
12:13 kowey joined #darcs
12:13 npouillard joined #darcs
12:13 gal_bolle joined #darcs
12:13 gh_ joined #darcs
12:13 Gowilla joined #darcs
12:13 lelit joined #darcs
12:13 darcswikibot joined #darcs
12:13 darcscommitbot joined #darcs
12:13 JaffaCake joined #darcs
12:13 Igloo_ joined #darcs
12:13 kpreid_ joined #darcs
12:13 gbeshers joined #darcs
12:13 sshc joined #darcs
12:13 Korusef joined #darcs
12:13 kolibrie joined #darcs
12:13 sm joined #darcs
12:13 thorkilnaur joined #darcs
12:13 nwf joined #darcs
12:13 dcoutts joined #darcs
12:13 lambdabot joined #darcs
12:13 idnar joined #darcs
12:13 mornfall joined #darcs
12:13 |Mike| joined #darcs
12:13 dleverton joined #darcs
12:13 kolmodin joined #darcs
12:13 karlpinc joined #darcs
12:13 PHP-Pierre_ joined #darcs
12:13 gwern joined #darcs
12:13 C-Keen joined #darcs
12:13 Jaak joined #darcs
12:13 felipe joined #darcs
12:13 tmug joined #darcs
12:13 dons joined #darcs
12:13 kosmikus joined #darcs
13:14 gh_ joined #darcs
13:18 gh_ joined #darcs
13:21 kpreid_ joined #darcs
13:37 lelit kowey: maybe you already know, reST (docutils) has a "csvtable" directive...
13:37 lelit able to load from an external file
13:38 tmug_ joined #darcs
13:38 kolmodin_ joined #darcs
13:38 dons joined #darcs
13:38 kosmikus joined #darcs
13:39 lelit http://docutils.sourceforge.net/​docs/ref/rst/directives.html#id1
13:43 kowey lelit: looks interesting (we could just turn on the CSV renderer from our tabular library)
13:43 kowey it'd have to be something that works with darcsit/pandoc in the short term, though
13:44 kowey but the external file loading does sound handy, since people could just upload into directories, eg. Benchmarking/Jason
13:45 kowey pandoc is awesome because it converts everything to everything -- only problem is that it can be lowest-common-denominator
13:46 lelit does pandoc implement the directives syntax?
13:48 kowey I don't know, but I doubt it because whatever it parses has to fit into this model: http://github.com/kowey/pandoc/blob/​master/src/Text/Pandoc/Definition.hs
13:48 kowey (nb, I'm not actually a Pandoc hacker; I have a fork because John MacFarlane hasn't gotten around to moving it off SVN yet)
13:52 lelit well, the directives are orthogonal (if I get the term right) to the model: (most of them) they just process their body as they like, and then insert standard elements in the doctree
13:52 kowey oh hmm...
13:52 lelit in the worst case, using a "literal" block, when the directive isn't known
13:53 lelit so, they are just an helper here, a fancy "include" feature
13:53 * kowey skims http://github.com/kowey/pandoc/blob/m​aster/src/Text/Pandoc/Readers/RST.hs
13:55 kowey It looks like it supports a handful of directives
13:55 lelit yeah
13:55 kowey although I wonder if it's easy to thread through "read an external" file into the parser
13:55 lelit for targets for example
13:56 kowey (in a clean way, that is)... but adding CSVtable support should be quite easy
13:56 kowey (since Pandoc already models tables), and natively supporting RST tables too
13:56 lelit and in the meantime, just "produce" a literal block
13:56 kowey ah right
13:56 lelit ah, I missed that!
13:57 lelit didn't you say that tables are unsupported?
13:57 kowey I think I meant "nobody got around to writing the parser for tables yet" rather
13:57 lelit ah, I see
13:58 kowey but there is a backend to tables that works for Markdown at least http://gitit.johnmacfarlane.net/Help
13:58 kowey so, I think it's another case of patches (very) welcome
13:58 kowey that'd be quite handy for the Darcs wiki, for example
13:59 lelit that would the best way, IMHO: it would be simple enough writing a little python script that extract the data from all the tables in a document, and rewrite that as JSON, if the need arise
14:02 lelit I'm so sorry I'm not up to the task, this picture explains better than any words [of mine] my usual reaction at following Haskell code... 8-) http://kvardek-du.kerno.org/2010/01/how-​common-lisp-programmer-views-users.html
14:02 kowey :-)
14:02 kowey I'm a little bit scared of the task myself, but I think that's just a bad habit
14:03 kowey the right attitude is to say "oh, that can't be too hard; just a little parsec, throw in a sepBy here and there, and you're set"
14:03 lelit yeah
14:04 kowey you can see the extended-markdown table parser here: http://github.com/kowey/pandoc/blob/mas​ter/src/Text/Pandoc/Readers/Markdown.hs
14:04 lelit the very few patches I contributed to darcs were born exactly like that, indeed
14:07 kowey oh... oh dear! I think I must have gotten confused in skimming the RST documentation
14:07 lelit it does not seem to implement a "read from external file"
14:07 lelit uh?
14:07 kowey and thought that the example output for tables was the actual syntax for writing them
14:07 kowey silly me...
14:08 kowey if you look at http://docutils.sourceforge.net/do​cs/ref/rst/directives.html#tables you see three syntaxes for tables (of which the CSVtable you mentioned)
14:08 lelit yes
14:08 kowey none of which are like the tables in http://wiki.darcs.net/RosettaStone
14:08 lelit something markdown parser does too
14:09 kowey and I think I got it into my head that RST tables were like that, because I superficially glanced at the RST page and saw some "tables" in pre
14:09 lelit wait
14:09 lelit we are talking about the table *directives* here!
14:09 lelit but reST does parse also inline tables
14:09 kowey is there a native table format?
14:09 kowey like the rosetta stone one?
14:10 kowey oh yes there is
14:10 lelit http://docutils.sourceforge.net/d​ocs/user/rst/quickref.html#tables
14:10 lelit two of them
14:10 kowey then I wasn't being stupid then, just being stupid now (forgot that I was looking at a directives only page)
14:11 lelit as said, AFAICT the markdown parser you urled above implements two similar kinds of "table descriptions"
14:20 kowey here's the doc for John's extended markdown tables - http://johnmacfarlane.net/​pandoc/README.html#tables
14:56 kpreid_ joined #darcs
15:28 lispy joined #darcs
15:52 lispy Hello
15:52 lambdabot lispy: You have 1 new message. '/msg lambdabot @messages' to read it.
15:53 lispy kowey: pbcopy is nice but the relevance is lost on me
16:03 gh_ joined #darcs
16:04 kpreid_ joined #darcs
16:04 gh_ joined #darcs
16:07 gh_ joined #darcs
16:12 malcolmw joined #darcs
16:21 darcswikibot 27 Jan 16:19 - EMMS switched to git (Guillaume Hoffmann)
16:33 gh_ http://wiki.darcs.net/ProjectsUsingDarcs needs a serious cleanout
16:34 gh_ a good amount of projects left using darcs, others have dead links, others have links pointing to pornographic websites
16:34 gh_ what about we don't try to list every project of the world using darcs, and keep the most visible? and remove the "Projects that stopped using darcs:" section
16:34 gh_ ?
16:43 kowey +1 this isn't 2003 anymore
16:44 kowey hi lispy: I was referring to the tables output... I speculated that the reason for poor formatting was copy+pasting from terminal to gmail
16:44 kowey (perhaps wrongly!)
16:44 gh_ working on it
16:44 lispy kowey: ah.  I think email clients do really weird things with line wrapping
16:44 kowey and that maybe pbcopy would avoid the problem of line breaks
16:44 lispy kowey: When in doubt, I send data in a format that is opaque to email clients
16:44 kowey even the list archives have funny wrapping, so I don't think it's on the reader's end
16:45 kowey oh, I think it's safe to drop 2.2.x (or at least 2.2.0) from the benchmarks now (in case this makes things faster for ya)
16:45 lispy kowey: you never know when 1) the sender's client will do something to it, 2) one of the mail servers in the chain of relays will reformat it, 3) when the recipient's end will reformat it
16:47 lispy the angband project is considering switching to dvcs so I'm sending them an email detailing why darcs is good and when it is not applicable
16:47 kowey angband?
16:47 * kowey googles
16:50 kowey lispy: yeah :-( re the difficulty of tracking down these formatting issues
16:50 lispy http://angband.oook.cz/pipermail/an​gband-dev/2010-January/000897.html
16:50 kowey that said, http://lists.osuosl.org/pipermail/d​arcs-users/2010-January/022833.html <-- see, it can work!
16:50 kowey :-D
16:51 lispy kowey: yeah, one constraint too is that the RFC makes some comments about lines over length 100
16:51 lispy I'm guessing my email had lines that wide
16:52 kowey (and if you have any patience for this, one thing to try maybe is to switch your gmail compose interface to plain text when transmitting these tables)
16:52 kowey (I don't want to make this benchmarking stuff any more of a pain than it already is, mind you)
16:52 kowey (I'm just saying, it may be worth trying out if you do produce future benchmarks)
16:52 kowey (I think I'll stop using these parentheses now)
16:53 lispy I honestly believe the problem this time was the width of the table.  It wasn't an issue with my last table, was it?
16:53 kowey well, that url above is a reply to one of your tables
16:53 kowey so I don't think width is necessarily the problem
16:53 kowey http://lists.osuosl.org/pipermail/d​arcs-users/2010-January/022832.html
16:53 kowey for the reply, I manually joined the lines in my text editor
16:54 lispy Oh, interesting.  so in my email client those lines looked fine but in the archive, not so much
16:54 lispy Yeah, I'll probably continue to use attachments then
16:54 lispy at least those will work
16:55 kowey OK
17:02 darcswikibot 27 Jan 17:01 - cleanup of projects using darcs (Guillaume Hoffmann)
17:06 kowey one idea might be to break this down by number of commits
17:06 kowey ballpark
17:08 kowey lispy: whoops, you accidentally said 'interactive gui'
17:08 lispy oh
17:08 lispy instead of ui
17:08 kowey but yeah, nice friendly, non-frothing-at-mouth mail :-)
17:09 kowey the fact that they're switching to *a* dvcs just makes life easier for everybody in any case
17:11 lispy Right
17:12 kowey I don't mean to insist, but I've found more evidence for my "blame rich text mode" hypothesis :-)
17:12 kowey which is that your mail actually comes in two flavours, one plain-text and one HTML
17:13 kowey and that the HTML one is just fine, so my working guess is that gmail's rich text => plain text converter hard-wraps things to be friendly
17:13 kowey anyway, it's no big deal, I just curious about these things sometimes!
17:14 lispy kowey: Well, I've had this issues historically irrespective of gmail
17:14 lispy Like, before I gmail existed
17:14 lispy "I knew..."
17:15 kowey ah, but have you ever been one of those grumpy people with anti-rich-text racism in their veins?
17:16 kowey anyway, I hereby drop the matter ... times have changed, rich text is here to stay, nothing I can do about it
17:17 darcswikibot 27 Jan 17:09 - forgotten ones (Guillaume Hoffmann)
17:18 lispy kowey: I used to be annoyed at MS versions of rich text things that were closed, proprietary, or broken; but these days rich-text can be "okay".
17:18 * kowey sniffs haughtily
17:18 kowey although I do confess to making very liberal use of rich text in Wave
17:41 darcswikibot 27 Jan 17:39 - merge Quickstart into GettingStarted, say more about tags (Guillaume Hoffmann)
17:43 gh_ bye
17:55 kpreid_ joined #darcs
18:21 arjanb joined #darcs
19:26 kpreid_ joined #darcs
19:39 gwern joined #darcs
19:56 kowey joined #darcs
20:13 balor joined #darcs
20:55 arjanb joined #darcs
21:18 kowey joined #darcs
23:02 kpreid_ joined #darcs
23:30 darcswikibot 27 Jan 23:29 - toc leveling ()
23:54 kpreid_ joined #darcs

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