MoinMoin: MoinMoinChat/Logs/moin-dev/2008-06-09

2008-06-09T00:03:07  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3684:f976d490b9df 1.7/MoinMoin/i18n/tools/po2wiki.py: use new ACLs / comments also when processing MoinI18n/xx pages
2008-06-09T00:06:16  <mitsuhiko> zenhase: does moin has a session system now?
2008-06-09T00:06:43  <xorAxAx> mitsuhiko: of course
2008-06-09T00:06:58  <mitsuhiko> when was that introduced?
2008-06-09T00:07:17  <ThomasWaldmann> 1.6
2008-06-09T00:07:22  <mitsuhiko> ah
2008-06-09T00:08:28  <xorAxAx> you can easily spot the session system in practice - its active when moin just throws away your page when saving in case of an outdated session :)
2008-06-09T01:18:26  <zenhase> re
2008-06-09T04:26:04  *** irc.freenode.net sets mode: +o ChanServ
2008-06-09T08:35:41  <ThomasWaldmann> moin
2008-06-09T08:51:59  <tuxella> hello ThomasWaldman
2008-06-09T08:56:12  <ThomasWaldmann> hi tuxella
2008-06-09T08:56:56  <ThomasWaldmann> you read about the pdf filter?
2008-06-09T09:16:58  <dreimark> oin
2008-06-09T09:18:04  <dreimark> m
2008-06-09T09:23:08  <tuxella> I only read the post on the wiki
2008-06-09T09:44:46  <dreimark> hi byeongweon
2008-06-09T10:57:09  <johill> ThomasWaldmann: ok, noted
2008-06-09T10:58:32  <johill> umm
2008-06-09T10:58:37  <johill> now I got a new user notification twice again
2008-06-09T10:58:51  <johill> I wonder if it has anything to do with the browser language being non-default
2008-06-09T11:01:35  <dennda> sorry for being a bit silent these last days, I got an exam dropping by soon (chose to write it now rather than in the "critical phase" of SoC)
2008-06-09T11:06:53  <zenhase> moin
2008-06-09T11:31:46  * dreimark lunch 
2008-06-09T12:29:28  <ThomasWaldmann> hmm, the sub/superscript parser/formatter is nuts
2008-06-09T12:59:10  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3685:56e6073b2e23 1.7/ (MoinMoin/parser/text_moin_wiki.py docs/CHANGES): Fix subscript parsing (was broken for cases like 'a,,1,2,,')
2008-06-09T13:25:42  <ThomasWaldmann> dreimark: you don't see the bug because i fixed it. you can still see it on the screenshot below.
2008-06-09T13:26:23  <dreimark> I meant the - box, will clarify it
2008-06-09T13:27:57  <dreimark> IE is doing some more strange
2008-06-09T13:28:45  <ThomasWaldmann> could be some "long" version of the minus not present in his charset
2008-06-09T13:46:44  <TheSheep_> -, –, — or ­?
2008-06-09T13:48:04  <ThomasWaldmann> http://moinmo.in/MoinMoinBugs/SubscriptRenderingIsBroken see screenshot there
2008-06-09T13:51:27  <TheSheep_> ThomasWaldmann: the minus sign in text is a normal "minus/dash" character, obviously can't tell anything about the screenshot
2008-06-09T13:52:47  <TheSheep_> ah, no, sorry
2008-06-09T13:52:58  <TheSheep_> it's the "real" minus sign from unicode
2008-06-09T13:53:01  <TheSheep_> it's not in ascii
2008-06-09T14:15:19  <ThomasWaldmann> well, client problem, not moin :)
2008-06-09T14:17:25  <TheSheep_> can we please "fix" all the ie-related bugs like this?
2008-06-09T14:32:42  <ThomasWaldmann> TheSheep_: hehe
2008-06-09T15:32:24  <johill> ThomasWaldmann: that looks like a strange workaround for a xapian query creator deficiency to me
2008-06-09T15:34:51  <johill> ThomasWaldmann: it looks like there should be a NotExpression doing that instead of AndExpression looking into the terms to see if they start with -
2008-06-09T15:36:10  <johill> ThomasWaldmann: ah, no. BaseExpression should learn that if it is negated, it needs can be xapian-enhanced... ahrg. visitor pattern would make this a lot simpler
2008-06-09T15:37:12  <johill> ThomasWaldmann: this is a stupid workaround, imho, but oh well, it doesn't matter all that much... I hope we can rewrite the xapian search properly anyway
2008-06-09T15:59:14  <xorAxAx> johill: i found a new model that might be interesting - "release consistency" :)
2008-06-09T16:00:14  <TheSheep> consi-what?
2008-06-09T16:01:06  <TheSheep> stenchy?
2008-06-09T16:01:10  <TheSheep> :)
2008-06-09T16:08:51  <johill> xorAxAx: the wikipedia description doesn't make a lot of sense to me
2008-06-09T16:12:01  <johill> xorAxAx: readers do not seem to 'acquire' the object, so what happens there? but if I understand it correctly, we're implementing 'eager release consistency'?
2008-06-09T17:07:53  <ThomasWaldmann> johill: yeah, a pure NOT would be nice, but I can't change it, xapian just can do a AND_NOT (and a MatchAll - and even that is rather new)
2008-06-09T17:08:56  <ThomasWaldmann> johill: so we just fix the tests to test for deeper nested stuff or do you see another easy way?
2008-06-09T17:08:58  <xorAxAx> johill: no idea what "we" are implementing, is there a wiki page?
2008-06-09T17:09:32  <johill> ThomasWaldmann: yeah, but the moin code need not express that xapian restriction, at least in theory
2008-06-09T17:10:08  <johill> xorAxAx: ask dennda? we decided that all the transaction business just leads to upper level code having to do retries all the time so actual locking is more sensible
2008-06-09T17:11:02  <ThomasWaldmann> ok, so I can either skip or adapt the tests
2008-06-09T17:11:53  <ThomasWaldmann> but moinmo.in with xapian is quite snappy :)
2008-06-09T17:11:57  <johill> well since the current xapian search requires the queries to be formed in a certain way, you should adapt them
2008-06-09T17:12:08  <xorAxAx> dennda: did you satisfy johill often occuring requirement of documenting design decisions?
2008-06-09T17:12:27  <johill> ThomasWaldmann: yeah I don't contest it, I just think that it's pretty stupidly designed
2008-06-09T17:12:46  <johill> xorAxAx: I think it's somewhere in the source code pretty much
2008-06-09T17:13:18  <johill> though I disagree with how it is currently called
2008-06-09T17:14:28  <ThomasWaldmann> I'll try again later today if I can make the AdvancedSearch form a bit more usable for normal users.
2008-06-09T17:15:26  <ThomasWaldmann> btw, the new search query parser obviously doesn't do what the advanced search stuff wanted it to do
2008-06-09T17:15:41  <ThomasWaldmann> it creates stuff like mimetype:foo,bar
2008-06-09T17:15:48  <johill> I didn't see anything about what it wanted so..
2008-06-09T17:16:00  <johill> hm
2008-06-09T17:16:10  <ThomasWaldmann> but that's easily fixed by doing (mimetye:foo or mimetype:bar)
2008-06-09T17:16:35  <johill> mimetypes probably cannot contain commas, but that syntax seems odd anyway
2008-06-09T17:16:42  <ThomasWaldmann> for categories, i guess we want AND (not OR)
2008-06-09T17:16:54  <ThomasWaldmann> and for languages maybe also OR
2008-06-09T17:18:15  <ThomasWaldmann> btw, the user search query stemming has a bit a problem currently for multilang sites
2008-06-09T17:18:52  <johill> not surprising
2008-06-09T17:19:35  <ThomasWaldmann> could be fixed by passing the language given by lang:xx to the right place in the code
2008-06-09T17:20:22  <johill> but what is the language?
2008-06-09T17:20:35  <johill> in a multilang wiki you don't really know what the user will be looking for
2008-06-09T17:20:38  <ThomasWaldmann> (but that would at the same time limit search to pages tagged with that language)
2008-06-09T17:21:22  <ThomasWaldmann> so if you have multilang pages, it won't work (well, stemming at index time also doesn't work for that)
2008-06-09T17:22:06  <ThomasWaldmann> not a big issues though as the analyzer creates (unstemmedword OR stemmedword) queries
2008-06-09T18:38:44  <dreimark> http://snippets.dzone.com/posts/show/5434
2008-06-09T18:38:52  <dreimark> for readable units
2008-06-09T18:53:49  <ThomasWaldmann> he either uses tabs or 8 spaces to indent O.O
2008-06-09T18:54:52  <ThomasWaldmann> and incorrect units :P
2008-06-09T18:55:54  <ThomasWaldmann> and 2 lines are completely superfluous
2008-06-09T19:12:26  * dreimark has seen the pep8 issues too,  with the fixed units and the right refactoring it can become nicer as formatInReadableUnits or _build_filelist
2008-06-09T19:13:17  <dreimark> btw. it is quite stupid to write tests and refactor code if you do this for data/plugin
2008-06-09T19:13:41  <dreimark> if you forget to update the testwiki you see no change
2008-06-09T19:14:30  <ThomasWaldmann> yes, but initializing a fresh testwiki for each test run is expensive
2008-06-09T19:15:23  <dreimark> that's true and surely not wanted
2008-06-09T19:17:27  <CIA-50> Reimar Bauer <rb.proj AT googlemail DOT com> default * 84:93125407c585 1.7-extensions/data/plugin/parser/text_x_arnica.py: text_x_arnica.create_thumbnail_and_webnail: bug fix does not try to create thumbnails from empty or corrupted image data file
2008-06-09T19:19:58  <johill> hopefully we can get rid of the test wiki with storage code
2008-06-09T19:21:42  <dreimark> i tried to call the test script from a test dir below data/plugin/parser/_tests
2008-06-09T19:21:45  <dreimark> moinpath =  os.path.abspath(os.path.join(os.path.dirname(sys.argv[1]), os.pardir, os.pardir, os.pardir, os.pardir, os.pardir))
2008-06-09T19:21:59  <dreimark> sys.path.insert(1, moinpath)
2008-06-09T19:22:11  <dreimark> that works for the import
2008-06-09T19:22:34  <dreimark> the current problem with that is that it does not get the request
2008-06-09T19:23:22  <dreimark> I don't have self.request
2008-06-09T19:23:44  <dreimark> any ideas ?
2008-06-09T19:26:01  <johill> nope, sorry
2008-06-09T19:35:49  <dreimark> bbl
2008-06-09T20:14:23  <CIA-50> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3686:e3b4ae470e86 1.7/MoinMoin/action/_tests/test_attachfile.py: test_attachfile: tests add_attachment with a mimetype file
2008-06-09T20:21:44  <ThomasWaldmann> wtf is a mimetype file?
2008-06-09T20:23:25  <dreimark> a png file encoded
2008-06-09T20:25:42  <dreimark> the test writes a png file and tests if it has the rigth size, should have added that to the description, sorry
2008-06-09T20:27:31  <ThomasWaldmann> well, what do you think makes this test important (compared to the one above it)?
2008-06-09T20:30:07  <dreimark> it tests saving in binary and it knows the length of the result
2008-06-09T20:30:20  <dreimark> and it is a small file
2008-06-09T20:30:25  <dreimark> < 1024
2008-06-09T20:32:06  <ThomasWaldmann> don't we always save binary data? it's not like that we have code handling attachment saving different depending on their content
2008-06-09T20:32:19  <ThomasWaldmann> and the other test is also a small file
2008-06-09T20:32:48  <ThomasWaldmann> (you could add the length test there)
2008-06-09T20:33:41  <dreimark> I think we should add the length test on many places.
2008-06-09T20:33:56  <ThomasWaldmann> if the point was testing against the bug one month ago with saving of small files, your test doesn't touch the critical parts of the code
2008-06-09T20:35:42  <dreimark> not only, I had choosen the size of the file small because of that previous bug.
2008-06-09T20:37:00  <dreimark> all the other tests do only save and test text, while we write binary
2008-06-09T20:39:37  <dreimark> bbl
2008-06-09T20:39:38  <ThomasWaldmann> a better test would be giving a file-like object as content btw.
2008-06-09T20:40:28  <ThomasWaldmann> (to test another code path than the test above)
2008-06-09T20:49:27  <dreimark> hmm, that was a real file beforehand I thought storing it base64 encoded is ok. it is a small black rectangel.
2008-06-09T20:58:06  <ThomasWaldmann> StringIO is file like without needing a on-disk file
2008-06-09T21:23:32  <dreimark> you think I should only test _write_stream in a string buffer ?
2008-06-09T21:31:32  <ThomasWaldmann> no, but if you add a second test for the same function, you should test another code path
2008-06-09T21:31:49  <ThomasWaldmann> by giving contents file-like not as str
2008-06-09T21:33:47  * dreimark is a bit confused
2008-06-09T21:35:22  <dreimark> I can read one of the images e.g. the logo and store this, but what should I do with StringIO, how does it help me ?
2008-06-09T21:40:23  <ThomasWaldmann> f = StringIO("somedata") creates a file-like object you can give to add_attachment as content
2008-06-09T21:45:17  <dreimark> ahh, I am thinking to complicated.  I wanted to replace the somedata word/text by the real data a image file has
2008-06-09T21:48:13  * dreimark walks home now
2008-06-09T21:48:31  <dreimark> bbl ~1h
2008-06-09T22:36:48  <dreimark> re
2008-06-09T23:12:13  <dreimark> ThomasWaldmann: http://paste.pocoo.org/show/66057/ ok?
2008-06-09T23:28:00  <dreimark> gn
2008-06-09T23:29:05  <ThomasWaldmann> much better
2008-06-09T23:29:19  <ThomasWaldmann> it has request as param?