Camelia, the Perl 6 bug

IRC log for #darcs, 2012-12-06

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

All times shown according to UTC.

Time Nick Message
00:00 kmels_ joined #darcs
00:20 schlaftier joined #darcs
00:27 mizu_no_oto joined #darcs
00:46 koninkje_away joined #darcs
01:02 carter_ joined #darcs
01:23 markstos joined #darcs
01:40 mizu_no_oto joined #darcs
02:04 intripoon_ joined #darcs
04:04 gbeshers joined #darcs
04:26 favonia joined #darcs
05:16 mizu_no_oto joined #darcs
07:01 intripoon joined #darcs
07:29 balor joined #darcs
08:20 drdo joined #darcs
08:42 raichoo joined #darcs
09:27 raichoo joined #darcs
09:30 balor joined #darcs
10:09 kmels joined #darcs
10:16 mndrix joined #darcs
10:37 owst joined #darcs
11:44 drostie joined #darcs
12:54 kmels joined #darcs
13:21 mizu_no_oto joined #darcs
13:36 gh_ joined #darcs
14:08 gbeshers joined #darcs
14:21 donri joined #darcs
14:36 iago joined #darcs
15:00 markstos joined #darcs
15:29 mizu_no_oto joined #darcs
15:54 favonia joined #darcs
16:18 raichoo joined #darcs
16:27 mizu_no_oto joined #darcs
16:45 iago joined #darcs
16:46 kmels_ joined #darcs
17:26 mizu_no_oto joined #darcs
17:52 schlaftier joined #darcs
18:23 kmels_ joined #darcs
18:25 kmels joined #darcs
18:30 amgarchIn9 joined #darcs
18:47 nomeata joined #darcs
18:50 iago joined #darcs
18:59 kmels_ joined #darcs
19:04 favonia joined #darcs
19:28 mizu_no_oto joined #darcs
19:48 markstos What's the difference between "darcs changes $file" and "darcs changes --only-to-files $file"?
19:48 markstos I compared the output of the two commands on a file with a long history of changes. "darcs changes --count" reported 152 patches in both cases.
19:49 markstos "--only-to-files" is not documented beyond the short phrase: http://darcs.net/manual/Darcs_comman​ds.html#SECTION00672000000000000000  "show only changes to specified files"
19:52 markstos I'm trying to research the history of a line in a file. "darcs annotate" is only good for the most recent changes, while "darcs changes --verbose filename" is close, there are identical looking lines in dozens of files, making the output confusing to sort through. Is there is a darcs option I'm missing for reviewing the history of a line?
19:57 balor joined #darcs
19:59 markstos This is helpful for my particular case: darcs show contents --match 'date "1 years ago"'
19:59 markstos I can use a binary search to narrow down when the output of this command matches a 'grep'
20:01 markstos wait, I thought there was a "darcs trackdown" command.
20:02 markstos Looks like "trackdown" was removed sometime after darcs 2.8?
20:08 * markstos tries to figure out why the Trackdown command is no longer present.
20:18 markstos Found it.   trackdown has been renamed to "test".  That should be highlighted in a Changelog if it's not already.  Perhaps I'll give it a whirl. (I was aware of trackdown for years, but always considered it too slow or cumbersome.  I'll see if --bisect changes that. )
20:21 markstos This is perhaps a further a optimization for "darcs test" that mesh well with the "Patch Index" work: Since sometimes people will know that the change must involve a particular file. Thus, all the algorithms could skip repo states which don't touch that particular touch. While my history has 18,000 patches in it, only about 150 involve changes to the file I'm researching.
20:25 markstos If it's not hard, it might be nice to leave "darcs trackdown" as a silent alias for "darcs test".
20:43 markstos My problem I'm finding today with "darcs test" is that I'm challenged to write the correct bash/grep syntax that does what I want: "test if this file does not contain this particular line".  I think it will be faster to manually experiment with "darcs show / darcs changes / darcs ann" than to sort that out. So, some examples of a few common test patterns used for "darcs test" would do a lot to enhance those docs, I think.
20:56 favonia joined #darcs
21:30 mizu_no_oto joined #darcs
21:45 Heffalump I think --bisect was actually in 2.8, btw. What's new in 2.9 is --backoff
21:46 markstos --backup was the one I ended up trying to day. It seemed the best fit for my case, and the performance was decent. It was constructing my test properly that was hard.
21:48 markstos "return true when grep doesn't find a match in this file, false when it does find a match" The Unix convention of return-0-on-success doesn't help  my conceptualization. :)
21:49 Heffalump :-)
21:54 Heffalump darcs annotate can be told to start from a certain patch, would that have helped?
21:55 markstos Interesting. Yes, I could have used the manual bisect method with that.
21:55 markstos I ended using this pattern: darcs show contents --match 'date "between 4 years ago and 2 years ago"' file.txt | grep foo.
21:56 markstos It performed pretty well, I could "back off" the ranges as I liked.
21:56 markstos It was a little confusing because I thought I was looking at a recently introduced issue, but it turned out to have been there since 2004,
21:57 markstos so the patch history was full mentions of it, often in giant blobs of HTML, because it was in a template. And when the template got updated, dozens of files changed and each time.
21:58 markstos I think what really happened is that very new version of Chrome just appear that started displaying our 2004 HTML differently.
22:15 delamonpansie joined #darcs
22:21 kmels joined #darcs
23:03 donri joined #darcs
23:10 stepcut joined #darcs

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