MoinMoin: MoinMoinChat/Logs/moin-dev/2008-08-10

2008-08-10T00:33:38  <ThomasWaldmann> xorAxAx: can it be that MoinClassCollector.setup runs twice, but test.setup_class only runs once?
2008-08-10T01:36:24  <dreimark> http://trac-hg.assembla.com/mercurialeclipse/wiki/Documentation
2008-08-10T01:36:30  <dreimark> cloning is slow
2008-08-10T01:42:12  <ThomasWaldmann> someone wake me up when mercurialeclipse starts working to an extent so one would rarely have to use hg cmdline
2008-08-10T01:44:57  <dreimark> ThomasWaldmann: it does not crash and it had done the clone
2008-08-10T01:48:22  <dreimark> annotate is cool
2008-08-10T02:08:48  * ThomasWaldmann fixes conftest
2008-08-10T02:19:23  * ThomasWaldmann slaps CIA-53 
2008-08-10T02:20:10  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3937:28442c70f757 1.8/MoinMoin/ (_tests/test_PageEditor.py conftest.py):
2008-08-10T02:20:10  <CIA-53> Fix conftest:
2008-08-10T02:20:10  <CIA-53> Do not only create a new request per test module, but per test class/function.
2008-08-10T02:20:10  <CIA-53> Fix one (broken) test.
2008-08-10T02:20:12  <CIA-53> Fix double creation of TestConfig (setup() is called twice by py.test, but
2008-08-10T02:20:14  <CIA-53> TestConfig must not get created twice.
2008-08-10T02:31:34  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4742:3cb7876135d4 1.8-storage-cdenter/MoinMoin/ (3 files in 2 dirs): merged moin/1.8
2008-08-10T02:31:35  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4743:984139f28deb 1.8-storage-cdenter/MoinMoin/ (Page.py security/__init__.py wikiutil.py): fix AccessControlList param data type
2008-08-10T02:32:40  <ThomasWaldmann> dennda: there is quite some other brokenness in the tests to fix
2008-08-10T02:33:05  <ThomasWaldmann> (or in the tested code :)
2008-08-10T02:43:20  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4716:d8ff2deedfad 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends_hg.py: (tests hg) Test fix (using non-public property)
2008-08-10T02:43:20  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4717:480f0c3c6c18 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py: (hg) Fix: _get_revision_metadata, _get_revision_timestamp. Clenup: imports, docstrings, comments.
2008-08-10T02:43:54  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4742:ef30768ddae3 1.8-mercurialbackend-ppacana/ (27 files in 11 dirs): Merge with 1.8-storage-cdenter.
2008-08-10T02:43:54  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4743:fcc3bb1e355a 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends.py: (backend tests) New tests: history, various revision metadata values. Updated test_timestamp.
2008-08-10T02:43:54  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4744:91a5c8b3a5f6 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/fs.py: (fs) Small history fix.
2008-08-10T02:44:29  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4745:703dc61ecfdd 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends_hg.py: (tests hg) Added comment.
2008-08-10T02:44:29  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4746:0d7f7e15c144 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py: (hg) Fixed metadata problem wrt hg internal conversion to str.
2008-08-10T02:46:36  <ThomasWaldmann> pawel: to use hg as clone src, you have to support reverse=False
2008-08-10T02:46:51  <pawel> i know
2008-08-10T02:47:14  * pawel is registered :/
2008-08-10T02:47:39  <ThomasWaldmann> that history code, is it taken from fs and modified?
2008-08-10T02:47:53  <PawelPacana> yes
2008-08-10T02:48:21  <ThomasWaldmann> then maybe first merge your changes back if possible and then just import from fs to share the code
2008-08-10T03:22:11  <PawelPacana> should be doable
2008-08-10T03:22:12  <PawelPacana> gn
2008-08-10T09:07:31  <CIA-53> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4744:9cbe6a72a2e5 1.8-storage-cdenter/tests/wikiconfig.py: storage: tests: Make Page tests work
2008-08-10T09:49:36  <CIA-53> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4745:96394af13789 1.8-storage-cdenter/MoinMoin/storage/backends/ (__init__.py fs17.py): storage: fs17-backend: Fixing metadata-values at the right place. (Thanks JohannesBerg)
2008-08-10T09:53:16  <xorAxAx> ThomasWaldmann: yes, possible. py.test has some serious bugs wrt the number of runs of different functions :-(
2008-08-10T10:05:22  <CIA-53> Byeongweon [tasyblue@gmail.com] default * 3988:ade975944b0d 1.8-guieditor-mbyeongweon/MoinMoin/ (5 files in 3 dirs): merged main
2008-08-10T10:05:26  <CIA-53> Byeongweon [tasyblue@gmail.com] default * 3989:9ab4430107e9 1.8-guieditor-mbyeongweon/MoinMoin/converter/text_html_text_moin_wiki.py: fix MoinMoinBugs/GuiEditorIndentingChangedForTables bug
2008-08-10T10:07:18  <xorAxAx> ThomasWaldmann: have you seen the google payment infos?
2008-08-10T10:13:33  <xorAxAx> moon: http://hg.moinmo.in/moin/1.8-guieditor-mbyeongweon/rev/9ab4430107e9 -- hmm, not sure whether that hack makes sense - i think the indenting is rather uniform
2008-08-10T10:15:55  <xorAxAx> moon: probably it works well, though, i will do some tests
2008-08-10T10:18:01  <xorAxAx> moon: there is another issue - i cannot indnet the table in the gui mode at all
2008-08-10T10:18:09  <xorAxAx> (thats a client side issue :))
2008-08-10T10:18:32  <xorAxAx> wow, opera segfault when using gui editor
2008-08-10T10:19:40  <moon> xorAxAx: I recognize that issue also. but it seems like not easy (means need more time).
2008-08-10T10:19:54  <moon> xorAxAx: opera? ok I wll try that.
2008-08-10T10:19:58  <xorAxAx> weee, reproducible segfault :-)
2008-08-10T10:20:05  <xorAxAx> well, its not your fault ...
2008-08-10T10:21:12  <moon> I think so. I just fix few js script these days. anyway, could you tell me reproduce step?
2008-08-10T10:21:40  <xorAxAx> i clicked into the PI section (grey bg), pressed enter a few times and changed the format to normal
2008-08-10T10:22:23  <xorAxAx> if you can reproduce, we can try to write shellcode and start our own botnet
2008-08-10T10:27:06  <moon> xorAxAx: hmm. is PI section means pre tag section?
2008-08-10T10:28:39  <xorAxAx> moon: no, you dont know what the PI section is? there are a few open bugs related to ti
2008-08-10T10:28:45  <xorAxAx> and i want to see them fixed :)
2008-08-10T10:29:44  <moon> actually. I don't know what is PI section extactly. could you tell me what it is?
2008-08-10T10:29:46  <xorAxAx> http://moinmo.in/HelpOnProcessingInstructions
2008-08-10T10:30:13  <moon> oh. is this PI section?
2008-08-10T10:30:19  <xorAxAx> yes
2008-08-10T10:30:34  <moon> hehe. I think PI related with math.
2008-08-10T10:30:36  <moon> :)
2008-08-10T10:30:53  <xorAxAx> heh, not many circles in moin there are
2008-08-10T10:33:35  <moon> hmm.. I think we should leave indenting table issue for the future. that has a lot of dependency with fckeditor main code.
2008-08-10T10:34:53  <moon> xorAxAx: do you know why we leave empty line on top of page? related with MoinMoinBugs/GUIEditorEatsEmptyLinesAtTop this bug.
2008-08-10T10:35:05  <xorAxAx> no idea
2008-08-10T10:36:28  <moon> that's side effect of useless line remove from converter.
2008-08-10T10:37:04  <xorAxAx> which line exactly?
2008-08-10T10:37:15  <xorAxAx> between the pi block end and the body start?
2008-08-10T10:38:03  <moon> line that mentioned to bug page?
2008-08-10T10:38:17  <moon> or ''useless line'?
2008-08-10T10:40:05  <moon> hmm.. I found same behavior on text editor also.
2008-08-10T10:42:28  <xorAxAx> ah
2008-08-10T10:42:41  <xorAxAx> well, why would anyone start a wikipage with empty lines?
2008-08-10T10:42:56  <xorAxAx> and want them to stay there
2008-08-10T10:44:48  <moon> xorAxAx: I don't know why. maybe dreimark knows reason.
2008-08-10T10:45:02  <xorAxAx> dreimark: use case for MoinMoinBugs/GUIEditorEatsEmptyLinesAtTop?
2008-08-10T10:45:53  <moon> we have meeting on this wednesday or monday?
2008-08-10T11:05:00  <dreimark> moin
2008-08-10T11:05:27  <dreimark> moon: how does it happen in the text editor, never seen there
2008-08-10T11:06:28  <dreimark> hmm I do see it in 1.7 now, probably that was changed sometime ago
2008-08-10T11:07:18  <moon> dreimark: in my chageset, insert empty line on top of page using text editor and save.
2008-08-10T11:07:37  <moon> and open with text mode I can't see empty line on top.
2008-08-10T11:07:48  <dreimark> moon: xorAxAx: no use case should only behave similiar to the text editor
2008-08-10T11:07:56  <dreimark> moon: enter two empty lines
2008-08-10T11:08:41  <dreimark> xorAxAx: moon the bug is on a different place
2008-08-10T11:09:02  <dreimark> every empty line should be removed if there is no reason to keep them
2008-08-10T11:09:31  <dreimark> otherwise the text editor reduces two empty lines to one and with the second saving to 0
2008-08-10T11:10:43  * dreimark extends the description
2008-08-10T11:15:43  <dreimark> moon: this bug is not related to your gui work, it is more generic
2008-08-10T11:44:02  <ThomasWaldmann>   moin
2008-08-10T12:01:48  <ThomasWaldmann> moon: please use list comprehensions instead of map/lambda
2008-08-10T12:27:13  <moon> ThomasWaldmann: ok. I will change such a code.
2008-08-10T12:33:20  <ThomasWaldmann> xorAxAx: yes
2008-08-10T13:04:42  <PawelPacana> moin
2008-08-10T13:05:04  <ThomasWaldmann> hi pawel
2008-08-10T13:15:41  <ThomasWaldmann> PawelPacana: that count array is not very readable
2008-08-10T13:17:26  <ThomasWaldmann> PawelPacana: s/compare_revision/same_revision/
2008-08-10T15:48:41  <dennda> PawelPacana: Why does clone() return copied revisions rather than copied items?
2008-08-10T15:48:46  <dennda> in count[0], that is
2008-08-10T15:54:47  *** irc.freenode.net sets mode: +o ChanServ
2008-08-10T15:58:17  <PawelPacana> basically because it iterating over revisions, and checking (on continued migration) if such revision exists in dest backend, thus if is converted
2008-08-10T15:58:44  <PawelPacana> if you need copied items count, i can change it to sth like skips or fails
2008-08-10T15:59:18  <PawelPacana> so you can len(converted.keys())
2008-08-10T16:30:47  <dennda> PawelPacana: The thing is that some tests rely on it returning the number of converted items
2008-08-10T16:42:00  <ThomasWaldmann> dennda: converted revisions is a more precise number, so maybe the test should rely on that?
2008-08-10T16:46:26  <xorAxAx> do items have to have at least one rev, dennda?
2008-08-10T16:46:55  <TheSheep> user storage doesn't have revisions at all
2008-08-10T16:47:03  <xorAxAx> so the answer is no :)
2008-08-10T16:47:06  <xorAxAx> and then ThomasWaldmann is wrong :)
2008-08-10T16:47:37  <TheSheep> xorAxAx: gasp!
2008-08-10T16:47:45  <ThomasWaldmann> no, I am right.
2008-08-10T16:47:51  <xorAxAx> (--> migrated 0 revisions is a bit useless in the user backend :))
2008-08-10T16:48:00  <ThomasWaldmann> but it should also test for item count. additionally.
2008-08-10T16:48:25  <xorAxAx> its independent - you can have more items than revisions
2008-08-10T16:49:33  <TheSheep> xorAxAx: but you cannot have revisions and no items
2008-08-10T16:50:00  <xorAxAx> TheSheep: but i could have one item with all revisions
2008-08-10T16:50:22  <TheSheep> lets find some more tautologies
2008-08-10T16:50:41  <TheSheep> xorAxAx: you could have one revision and three ites,
2008-08-10T16:50:41  <xorAxAx> hehe
2008-08-10T16:51:09  <xorAxAx> i also could have 3 backends with no items and a revision in a fourth one
2008-08-10T16:51:18  <TheSheep> xorAxAx: the question is, what's that you want to know?
2008-08-10T16:51:30  <xorAxAx> what to know from whom?
2008-08-10T16:51:39  <TheSheep> xorAxAx: from that function's return value
2008-08-10T16:52:05  <xorAxAx> i want to know an instance of a class :)
2008-08-10T16:52:19  <xorAxAx> with as much information as my user needs. user being the caller
2008-08-10T16:52:23  <TheSheep> ok, see, it wasn't that hard
2008-08-10T16:52:31  <TheSheep> PawelPacana: make it return an instace of a class
2008-08-10T16:52:34  <TheSheep> PawelPacana: any class
2008-08-10T16:52:35  <xorAxAx> :-)
2008-08-10T16:52:51  <PawelPacana> ;-)
2008-08-10T16:53:18  <xorAxAx> "Migration success: <object object at 0xb7da4468>"
2008-08-10T16:53:42  <TheSheep> so maybe it should return true on success and false on error?
2008-08-10T16:53:54  <PawelPacana> dennda: will be changed soon
2008-08-10T16:53:54  <TheSheep> or maybe it should return None on success and raise on error?
2008-08-10T16:54:09  <xorAxAx> what kind of error?
2008-08-10T16:54:29  <TheSheep> xorAxAx: oh, you can raise different exceptions for different errors
2008-08-10T16:54:58  <xorAxAx> sigh :)
2008-08-10T16:55:12  <xorAxAx> which errors would happen in the clone script that you want to see raised?
2008-08-10T16:58:13  <TheSheep> xorAxAx: hmm... failed revisions, probably in bulk
2008-08-10T16:58:55  <xorAxAx> if you always end up completling the whole loop, you wouldnt really raise but rather have an attr of the status class with some list of errors
2008-08-10T16:59:11  <xorAxAx> but i am not sure if we are overdesigning here
2008-08-10T16:59:42  <xorAxAx> also most errors will probably happen in the backend itself, leading to an exception
2008-08-10T17:00:31  <ThomasWaldmann> as we have to copy with some imperfection, the clone should not stop if it encounters problems with some item/rev, but just notify about it
2008-08-10T17:00:41  <ThomasWaldmann> s/copy/cope/
2008-08-10T17:01:26  <ThomasWaldmann> it is really annoying if you have to fix the source data one item per broken clone trial
2008-08-10T17:01:37  <TheSheep> yeah
2008-08-10T17:02:14  <ThomasWaldmann> the usual thing happening is crap metadata and missing revs
2008-08-10T17:02:31  <TheSheep> otoh the count of converted revs is only used internally, right?
2008-08-10T17:02:47  <TheSheep> in tests
2008-08-10T17:04:55  <xorAxAx> ThomasWaldmann: how do you suggest to proceed if the source backend raises an exception on certain ops? let the user know that the particular item/revs were dropped and that he should deal with it somehowß
2008-08-10T17:05:26  <xorAxAx> probably makes sense in all scenarios where an admin is involved at the prompt
2008-08-10T17:06:27  <ThomasWaldmann> xorAxAx: yes. it should output a list of problems.
2008-08-10T17:06:35  <ThomasWaldmann> then admin can fix all of them at once.
2008-08-10T17:09:53  <dennda> doesn't it sound more sane to just fix the errors where they come from, that is, the source backend?
2008-08-10T17:10:05  * dennda is washing the dishes, so semi afk
2008-08-10T17:10:19  <dennda> PawelPacana: thanks
2008-08-10T17:11:12  <PawelPacana> dennda: done
2008-08-10T17:11:24  *** irc.freenode.net sets mode: +o ChanServ
2008-08-10T17:11:25  <PawelPacana> take len() of first returned param
2008-08-10T17:11:41  <ThomasWaldmann> dennda: sure, but some stuff is hard to fix. there could be some heuristics, though.
2008-08-10T17:11:58  <dennda> ThomasWaldmann: I don't know, *how* does it fail?
2008-08-10T17:12:30  <ThomasWaldmann> e.g. if your edit-logs are crap because of NFS problems or moin coding errors, then data you need is just not there.
2008-08-10T17:13:03  <dennda> how do you handle that?
2008-08-10T17:13:18  <ThomasWaldmann> fs17 has some code to just make something up
2008-08-10T17:13:27  <dennda> "just make something up"?
2008-08-10T17:14:25  <ThomasWaldmann> if you have no metadata, but you technically require it, you can put empty strings, fake IPs, etc. in it
2008-08-10T17:15:22  <ThomasWaldmann> the biggest problem related to this yet, has been the timestamp. currently it puts 0 there, but of course that generates wrong history() order
2008-08-10T17:15:36  <xorAxAx> OMG, wrong history!
2008-08-10T17:15:49  <xorAxAx> ThomasWaldmann: well, then it needs to return None probably
2008-08-10T17:16:06  <ThomasWaldmann> could be improved by looking at filesystem mtime and other edit-log entries
2008-08-10T17:17:08  <xorAxAx> hmm, none is also bad
2008-08-10T17:17:13  <ThomasWaldmann> xorAxAx: it's not so much about "OMG wrong history", but early versions of clone just falling over that
2008-08-10T17:17:14  <xorAxAx> because clone cant interpolate it
2008-08-10T17:17:28  <xorAxAx> ThomasWaldmann: well, if history() is wrong, other parts will fall over it :)
2008-08-10T17:17:36  <dennda> and shouldn't, imho
2008-08-10T17:17:52  <xorAxAx> its not easy
2008-08-10T17:19:15  <ThomasWaldmann> for now, I am happy with a clone converting everything that doesn't error and make a list of the problems
2008-08-10T17:21:03  <ThomasWaldmann> RecentChanges still crashes silently when just run on converted underlay data
2008-08-10T17:23:35  <ThomasWaldmann> Es wurden keine Anhänge für die Seite TitelIndex gespeichert.  (action=AttachFile)
2008-08-10T17:24:04  <ThomasWaldmann> Es sind 120 Anhänge für diese Seite gespeichert. (action=info&general=1 of same page)
2008-08-10T17:24:30  <dennda> ls -la /tmp | wc -l
2008-08-10T17:25:15  <TheSheep> 65536
2008-08-10T17:25:36  <TheSheep> dennda: why -l ?
2008-08-10T17:25:43  <TheSheep> dennda: in ls
2008-08-10T17:25:44  <dennda> I want the number of lines
2008-08-10T17:26:28  <ThomasWaldmann> action=info&hitcounts=1 tells something about 1970-01-01
2008-08-10T17:26:45  <dennda> which is probably where mtime started
2008-08-10T17:26:48  <dennda> if you fill in 0
2008-08-10T17:27:49  <ThomasWaldmann> it is listing 4 accesses and my system clock is not in the 70s X)
2008-08-10T17:28:47  <dennda> ?
2008-08-10T17:29:14  <ThomasWaldmann> i read that page 4 times
2008-08-10T17:29:56  * ThomasWaldmann runs WantedPages
2008-08-10T17:30:28  <ThomasWaldmann> btw, I'll fix the wrong "has attachments" stuff on TitleIndex by just removing that feature
2008-08-10T17:36:13  <ThomasWaldmann> heh, PageSize listing now has pages with size 0
2008-08-10T17:36:32  <ThomasWaldmann> (because the #redirect on them is now in metadata)
2008-08-10T17:37:39  <dennda> ThomasWaldmann: sorry I don't get it
2008-08-10T17:38:13  <dennda> the strange date behaviour is completely expected if you fill in bogus metadata
2008-08-10T17:38:54  <ThomasWaldmann> the 4 hits are read accesses and thus completely unrelated to the page rev timestamp
2008-08-10T17:39:17  <ThomasWaldmann> 69 if revbefore and page.exists(rev=revbefore, domain='standard'):   < traceback because of rev param
2008-08-10T17:39:29  <dennda> if you read the page 4 times, what is wrong with it displaying it has been accessed 4 times?
2008-08-10T17:39:36  <ThomasWaldmann> (RecentChanges)
2008-08-10T17:39:50  <ThomasWaldmann> it tells something about 1970
2008-08-10T17:40:00  <dennda> of course it does
2008-08-10T17:40:07  <dennda> that is because you set the timestamp to 0
2008-08-10T17:40:18  <dennda> unix time starts that day
2008-08-10T17:40:36  <ThomasWaldmann> maybe just look at the RecentChanges thing
2008-08-10T17:41:10  * ThomasWaldmann can live with broken stats for now, but not with broken RC
2008-08-10T17:42:45  <PawelPacana> bbl
2008-08-10T17:43:15  <ThomasWaldmann> full text search is also broken. try fts for wiki ...
2008-08-10T17:45:07  <ThomasWaldmann> WantedPages ended with "too many files open"...
2008-08-10T17:49:45  <ThomasWaldmann> AttachFile has some problems still
2008-08-10T17:54:28  * ThomasWaldmann slaps CIA-54 
2008-08-10T18:01:54  <ThomasWaldmann> dennda: what are you working on currently (/me wants to avoid conflicts)?
2008-08-10T18:02:12  <dennda> attachfile
2008-08-10T18:02:43  <ThomasWaldmann> ok, I guess I try killing getPageDict now
2008-08-10T18:25:33  <ThomasWaldmann> woah, fixing WantedPages opens a can of worms :|
2008-08-10T19:23:49  <ThomasWaldmann> ok, getPageDict removed
2008-08-10T19:27:17  <ThomasWaldmann> dennda: still at attachfile?
2008-08-10T19:36:50  * ThomasWaldmann changes caching.CacheEntry.needsUpdate for new requirements
2008-08-10T19:59:10  <dennda> sorry, was away a while, housekeeping and such
2008-08-10T20:22:53  <ThomasWaldmann> yeah, teh src also needs some housekeeping :D
2008-08-10T20:23:11  <dennda> it's a little mess, yes
2008-08-10T20:24:04  <dennda> but if the maggots devour me there'll be no one to clean it up %-)
2008-08-10T20:42:45  <ThomasWaldmann> ok, CacheEntry.needsUpdate fixed, Page._text_filename removed
2008-08-10T20:54:49  <dennda> how many tests fail for you now?
2008-08-10T21:03:04  <ThomasWaldmann> 6 or 7
2008-08-10T21:03:18  <ThomasWaldmann> some of them you could fix rather easily
2008-08-10T21:03:24  <ThomasWaldmann> dennda:
2008-08-10T21:03:38  <dennda> ok
2008-08-10T21:07:44  <dennda> redundancies all over the place
2008-08-10T21:11:26  <CIA-54> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4752:9a17a1b3b947 1.8-storage-cdenter/MoinMoin/action/AttachFile.py: storage: action/AttachFile: Make view basically work
2008-08-10T21:11:27  <CIA-54> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4753:890fe215ec50 1.8-storage-cdenter/MoinMoin/ (Page.py _tests/test_caching.py caching.py i18n/__init__.py): Merge my own branch, 1.8-storage-cdenter
2008-08-10T21:13:09  <dennda> sigh, found a bug in 1.8... fixing...
2008-08-10T21:22:20  <CIA-54> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4754:aa5f48d743fc 1.8-storage-cdenter/MoinMoin/search/term.py: fix FromUnderlay term
2008-08-10T21:22:21  <CIA-54> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4755:afe1acb4681e 1.8-storage-cdenter/MoinMoin/action/AttachFile.py: merge storage repo
2008-08-10T21:24:45  <CIA-54> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4756:b28f5f17b09b 1.8-storage-cdenter/MoinMoin/action/AttachFile.py: storage: action/AttachFile.py: When no filename is specified in the attachment-upload-form, don't display a NoneTypeError. Adding myself to Copyright-Section.
2008-08-10T21:24:46  <CIA-54> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 4757:d4a89a7470d0 1.8-storage-cdenter/MoinMoin/search/term.py: Merge my own repo, 1.8-storage-cdenter
2008-08-10T21:24:50  <dennda> ThomasWaldmann: ^^^
2008-08-10T21:24:57  <dennda> That bug is also present in 1.8-main
2008-08-10T21:39:10  <ThomasWaldmann> dennda: that patch could trigger problems
2008-08-10T21:39:21  <dennda> elaborate please
2008-08-10T21:39:45  <ThomasWaldmann> iirc, not every server type provides that __filename stuff
2008-08-10T21:40:21  <ThomasWaldmann> in those cases, we required the rename field to be filled out
2008-08-10T21:41:02  <dennda> Well, I didn't write that part of the code so I don't know. As far as I am concerned, that if statement was at the wrong place
2008-08-10T21:41:14  <dennda> you got another bug then
2008-08-10T21:42:29  <ThomasWaldmann> i think we don't need change there right now. upload stuff is going to get handled differently anyway.
2008-08-10T21:44:31  <dennda> well, just don't forget it then
2008-08-10T21:46:50  <ThomasWaldmann> dennda: how's mimetype metadata handled right now in your repo?
2008-08-10T21:48:08  <dennda> "mimetype" metadata key
2008-08-10T21:52:54  <ThomasWaldmann> ehrm, "unknown" is no valid mimetype
2008-08-10T21:53:46  <ThomasWaldmann> maybe you should rather use application/octet-stream for that
2008-08-10T21:54:45  <ThomasWaldmann> similar for this x-unidentified-wiki-format. just do it right.
2008-08-10T21:57:42  <ThomasWaldmann> creation of Terms that match stuff that is considered as "attachment" like stuff as well as for Terms that match stuff considered as "page" should be added where the other Terms are defined
2008-08-10T21:59:05  <dennda> ThomasWaldmann: Could you rephrase the last sentence or put a few commas or something in there? Difficult to understand... :-)
2008-08-10T22:00:18  <ThomasWaldmann> in AttachFile there is some search term constructed to match stuff considered as attachment
2008-08-10T22:00:30  <ThomasWaldmann> line 436
2008-08-10T22:00:52  <dennda> yes
2008-08-10T22:00:55  <ThomasWaldmann> that part of the term should be available separately
2008-08-10T22:01:03  <ThomasWaldmann> and also its negation
2008-08-10T22:01:10  <dennda> some constant?
2008-08-10T22:01:42  <ThomasWaldmann> like the underlay term
2008-08-10T22:01:54  <dennda> underlay term?
2008-08-10T22:02:06  <ThomasWaldmann> search.term
2008-08-10T22:03:30  <dennda> yes, but where is that underlay term defined?
2008-08-10T22:04:27  <ThomasWaldmann> at the end of that module
2008-08-10T22:04:53  <ThomasWaldmann> just create a Term that matches on pages and one that matches attachments
2008-08-10T22:05:19  <ThomasWaldmann> (of course it is all items now, but you can look at the current rev's mimetype)
2008-08-10T22:06:00  <ThomasWaldmann> you need a list of what can be a wiki page. the attachments Term is just NOT(pageterm)
2008-08-10T22:06:24  <dennda> I would rather use a seperate term for that
2008-08-10T22:06:45  <dennda> users are neither pages nor attachments (they are defined in a different backend, but still)
2008-08-10T22:07:06  <dennda> but I think we can do it either way
2008-08-10T22:07:23  <ThomasWaldmann> you don't need to care for users because that is a different backend
2008-08-10T22:07:35  <dennda> yes, I just said that :)
2008-08-10T22:07:49  <ThomasWaldmann> and you can't enlist all mimetypes that are NOT pages
2008-08-10T22:08:34  <dennda> thus NOT(LastRevisionMetaDataMatch('mimetype', 'text/x-unidentified-wiki-format')),
2008-08-10T22:09:16  <ThomasWaldmann> no
2008-08-10T22:09:38  <ThomasWaldmann> indentify first
2008-08-10T22:10:06  <dennda> so, in a nutshell: What do you want me to do exactly?
2008-08-10T22:10:08  <dennda> ?
2008-08-10T22:10:15  <dennda> with the regex?
2008-08-10T22:10:27  <ThomasWaldmann> you remove that crap x-unidenti... first
2008-08-10T22:10:45  <ThomasWaldmann> then you use real mimetypes
2008-08-10T22:11:02  <dennda> Why is that not a real mimetype?
2008-08-10T22:11:04  <ThomasWaldmann> and then you will have a list of mimetypes considered as a page
2008-08-10T22:11:17  <dennda> Is there such a list?
2008-08-10T22:11:37  <ThomasWaldmann> it's useless, we need something so we can differentiate between wiki, rst, ...
2008-08-10T22:12:36  <dennda> well, ok, but I will need a list of mimetypes that represent pages
2008-08-10T22:12:58  <ThomasWaldmann> and i don't care much right now whether you use text/x-foo or text/foo as long as it is consistent within moin source
2008-08-10T22:13:23  <ThomasWaldmann> well, create that list :)
2008-08-10T22:14:21  <dennda> You know the possible mimetypes much better than me, I can't just guess
2008-08-10T22:14:32  <ThomasWaldmann> having a look into parser/* might help
2008-08-10T22:15:45  <ThomasWaldmann> there is also wikiutil.MimeType
2008-08-10T22:21:55  <ThomasWaldmann> btw, mimetype metadata is new, the older thing we used was just #format
2008-08-10T22:22:38  <ThomasWaldmann> but format should go away and get replaced by mimetype imho
2008-08-10T22:22:39  <dennda> Yeah, maybe I removed that already a while back when I started with AttachFile
2008-08-10T22:23:06  <ThomasWaldmann> AttachFile never had any mimetype/format metadata
2008-08-10T22:24:45  <dennda> Can't recall
2008-08-10T22:25:17  <ThomasWaldmann> there was simply no place where to store it, it was just the file and the edit-log
2008-08-10T22:27:28  <dennda> then that is what I removed :)
2008-08-10T22:27:52  <dennda> My memory is like a sieve
2008-08-10T22:28:24  <ThomasWaldmann> (btw, until this stuff is solved, getPageList is rather broken, because it returns all items, not just the pages. if then someone calls getPageLinks on such a "page", weird stuff happens, if that is just binary data...)
2008-08-10T22:30:08  <ThomasWaldmann> getPageList could just be a wrapper around getItemList telling that it just wants "pages"
2008-08-10T22:34:56  <xorAxAx> well, here the first problems begin
2008-08-10T22:35:11  <xorAxAx> if you manage to store an item somewhere where it is not an "attachment" it will be "gone"
2008-08-10T22:35:30  <xorAxAx> so you should be probably stopped to do so
2008-08-10T22:36:02  <ThomasWaldmann> hm?
2008-08-10T22:36:57  <TheSheep> ThomasWaldmann: shouldn't getPageLinks check the type anyways, to know which parser to use?
2008-08-10T22:37:38  * ThomasWaldmann looks
2008-08-10T22:38:10  <xorAxAx> ThomasWaldmann: well, just add an item /foo.mp3
2008-08-10T22:38:32  <xorAxAx> ThomasWaldmann: and your wiki starts being an mp3 exchange place and it might not even appear e.g. on titleindex
2008-08-10T22:39:04  <TheSheep> xorAxAx: it appears on recent changes
2008-08-10T22:39:11  <ThomasWaldmann> xorAxAx: it just depends on mimetype, not on storage place
2008-08-10T22:39:12  <TheSheep> xorAxAx: isn't that enough for peer review?
2008-08-10T22:39:37  <xorAxAx> TheSheep: probably
2008-08-10T22:39:44  <ThomasWaldmann> and if we want to see all "file like things" we just have to have ItemIndex or some option to TitleIndex
2008-08-10T22:40:23  <ThomasWaldmann> like we now can include underlay stuff or not, we could also include files or not
2008-08-10T22:43:25  <xorAxAx> yeah
2008-08-10T22:44:41  <ThomasWaldmann> TheSheep: it is because it is still using pi['format']
2008-08-10T22:45:29  <ThomasWaldmann> dennda: Page:1018
2008-08-10T22:46:45  <dennda> that should rather be rev["metadata"]?
2008-08-10T22:47:42  <dennda> @param format: format of content, default 'wiki'
2008-08-10T22:47:42  <moinBot> dennda: Error: "param" is not a valid command.
2008-08-10T22:48:53  * xorAxAx hugs moinBot 
2008-08-10T22:49:21  <xorAxAx> totally apathetic
2008-08-10T22:51:56  <TheSheep> when did we get a bot?
2008-08-10T22:52:34  <xorAxAx> 3 years ago
2008-08-10T22:52:55  <TheSheep> come to think of it, these logs have to come from somewhere...
2008-08-10T22:53:28  * TheSheep feels like those kids who never thought where the meat comes from
2008-08-10T22:54:47  <xorAxAx> surveillance is best when undetected
2008-08-10T22:57:10  <dennda> true
2008-08-10T22:57:24  <xorAxAx> omg
2008-08-10T23:09:02  <dennda> xorAxAx: hm?
2008-08-10T23:10:34  <xorAxAx> dennda: its not really true, is it? :) but no political discussions here, coding is much more rewarding! :)
2008-08-10T23:26:14  <ThomasWaldmann> >           ItemAlreadyExistsError: An Item with the name u'AutoCreatedMoinMoinTemporaryTestPageX' already exists!
2008-08-10T23:43:38  <ThomasWaldmann> waldi: 42 failures in your repo (and not only in macro tests)
2008-08-10T23:44:21  <birkenfeld> xorAxAx: shut up and code? I like that :)
2008-08-10T23:51:41  <xorAxAx> hehe
2008-08-10T23:51:51  <xorAxAx> its called refocusing :)