1 2010-04-02T00:07:01 *** pchen
2 2010-04-02T00:07:14 <pchen> hello
3 2010-04-02T00:07:25 <pchen> I'm getting an error when running /bin/moin
4 2010-04-02T00:07:44 <pchen> The error occurs when /MoinMoin/logfile/editlog.py is called by the moin script
5 2010-04-02T00:45:33 *** pchen
6 2010-04-02T00:50:17 *** valeuf_
7 2010-04-02T00:52:49 *** valeuf
8 2010-04-02T00:52:49 *** valeuf_
9 2010-04-02T04:42:35 *** AshishG
10 2010-04-02T06:02:18 *** moin0
11 2010-04-02T06:02:26 <moin0> hi
12 2010-04-02T06:02:43 *** moin0
13 2010-04-02T06:02:57 <yo_colombia> is too hard python?
14 2010-04-02T06:37:34 *** yo_colombia
15 2010-04-02T09:06:23 *** valeuf
16 2010-04-02T09:12:13 *** valeuf
17 2010-04-02T11:07:01 *** selevt
18 2010-04-02T15:42:01 *** JosefMeier
19 2010-04-02T15:42:18 <JosefMeier> Moin
20 2010-04-02T15:48:54 *** selevt_
21 2010-04-02T15:52:23 <JosefMeier> dreimark: Do you have news about the xapian probs in 1.9.2 ?
22 2010-04-02T15:52:25 *** selevt
23 2010-04-02T15:53:42 <dreimark> JosefMeier: i have a patch added to the page
24 2010-04-02T15:54:22 <dreimark> http://moinmo.in/MoinMoinBugs/1.9.2XapianRegexNeedsCase?action=AttachFile&do=view&target=xapian_re_search_opt.patch
25 2010-04-02T16:12:53 *** JosefMeier
26 2010-04-02T16:58:36 *** JosefMeier
27 2010-04-02T16:59:14 <JosefMeier> dreimark: Thanks for the info. I'll try that patch. Is this patch also installed on the official moin homepage?
28 2010-04-02T17:03:37 <dreimark> JosefMeier: not yet
29 2010-04-02T17:14:07 <ThomasWaldmann> re
30 2010-04-02T17:18:25 * ThomasWaldmann had a small chat with lfrodrigues about searching/indexing
31 2010-04-02T17:25:18 <ThomasWaldmann> btw, that xapian bug is still much too unclear / confused. cause of the problem is unclear and the patch is unclear also.
32 2010-04-02T17:28:45 <ThomasWaldmann> in general, one should be able to clearly identify the cause of a bug (== point to it in the source) before one starts "fixing" it
33 2010-04-02T17:35:25 <dreimark> the graphic shows exactly the problem we have nothing to search in
34 2010-04-02T17:37:02 *** amartani
35 2010-04-02T17:40:24 <dreimark> have a look at it, we have no content to search by xapian
36 2010-04-02T17:43:36 <ThomasWaldmann> (!) "case:" triggers post-processing of Xapian search results by MoinSearch (because xapian indexing/search is not case-sensitive). Strange: MoinSearch will get fed with a pagelist pre-selected by xapian search. So if xapian search finds nothing, post-processing should not help either.
37 2010-04-02T17:43:42 <ThomasWaldmann> still unanswered
38 2010-04-02T17:45:16 <JosefMeier> ThomasWaldmann, dreimark: Do you think that the bug is in xapian itself?
39 2010-04-02T17:45:24 <ThomasWaldmann> Because regex search uses match for comparison this will only give a result if there is a line or a page title given which starts with "bindi".
40 2010-04-02T17:45:31 <ThomasWaldmann> that sounds like another bug...
41 2010-04-02T17:49:57 <ThomasWaldmann> JosefMeier: likely no, our code is just not right
42 2010-04-02T17:52:33 <ThomasWaldmann> and pumping all wiki content (optionally even all page revisions) into the xapian index doesn't feel right either
43 2010-04-02T17:52:36 <dreimark> yes, and such stuff you find also only if you look at the code
44 2010-04-02T17:52:54 <dreimark> but what if this the only one solution?
45 2010-04-02T17:56:31 <dreimark> case search is selecting MoinSearch
46 2010-04-02T17:59:37 <dreimark> JosefMeier: i think i have not stopped the refactoring early enough. in the past we did a regex text search by moinsearch now we select the searcher and that searches then
47 2010-04-02T18:00:54 <dreimark> it tries to make a regex search in content of the xapian database
48 2010-04-02T18:01:23 <dreimark> but we don't have content there
49 2010-04-02T18:02:28 <dreimark> so the patch makes a storage backend with duplicated content
50 2010-04-02T18:03:49 <dreimark> and that can cost some space
51 2010-04-02T18:11:49 <dreimark> ThomasWaldmann: http://hg.moinmo.in/moin/1.9/annotate/689e2b04bd4d/MoinMoin/search/Xapian/search.py#l78
52 2010-04-02T18:12:02 <dreimark> i think this is the cause for the case results
53 2010-04-02T18:24:49 <dreimark> that line just makes not that post-processing it just did a search
54 2010-04-02T18:28:57 <dreimark> ThomasWaldmann: if we have nothing found by xapian pages is []
55 2010-04-02T18:29:01 <dreimark> and then http://hg.moinmo.in/moin/1.9/file/689e2b04bd4d/MoinMoin/search/builtin.py#l383
56 2010-04-02T18:29:17 <dreimark> makes a complete new search by the moinsearch
57 2010-04-02T18:29:38 <JosefMeier> dreimark: Is it possible to search in attachments also with your patch?
58 2010-04-02T18:30:56 <dreimark> just asking or found that i broke it or was it broken before too?
59 2010-04-02T18:34:00 <dreimark> JosefMeier: because the attachment search test looks not broken and does not fail after adding my patch it should work
60 2010-04-02T18:36:29 * dreimark looks again at the test
61 2010-04-02T18:39:28 <JosefMeier> dreimark: just asking
62 2010-04-02T18:41:13 *** amartani
63 2010-04-02T18:56:01 <dreimark> JosefMeier: if you find an other issue please create a new bug report.
64 2010-04-02T18:56:31 <dreimark> we don't fix all these bugs in one cs
65 2010-04-02T19:04:49 <JosefMeier> dreimark: ok. What's the result of your discussion with Thomas about your patch? Does it solve the problem or do you think there has more work to be done?
66 2010-04-02T19:12:06 <dreimark> if i look at the backlog we don't have a discussion, we both told lines when the other one was not online.
67 2010-04-02T19:12:56 <dreimark> a smarter solution would be nicer but i doubt that there is one
68 2010-04-02T19:13:56 <dreimark> also if xapian itselfs could handle regex search terms that would be the best solution, but currently it isn't
69 2010-04-02T19:14:10 <dreimark> possible
70 2010-04-02T19:30:34 <ThomasWaldmann> dreimark: space is not the only issue with that approach
71 2010-04-02T19:31:53 <ThomasWaldmann> but first doing a regex search over all content just to build potentially huge ORed xapian expressions that finally will find the same stuff again...
72 2010-04-02T19:37:37 <dreimark> that is quite expensive
73 2010-04-02T19:41:16 <ThomasWaldmann> that is quite insane :)
74 2010-04-02T19:41:49 <dreimark> but it is supposed to have these xapian expressions
75 2010-04-02T19:42:14 <ThomasWaldmann> for some rather limited amount of data (like item names), that's doable, but for all content...
76 2010-04-02T19:42:51 <dreimark> i am sometimes not sure if one has more content by attachments than by pages
77 2010-04-02T19:43:09 <ThomasWaldmann> that doesn't matter
78 2010-04-02T19:43:55 <dreimark> that also could have bid xapian expressions
79 2010-04-02T19:44:13 <dreimark> it just not have the content regex search beforhand
80 2010-04-02T19:48:22 <ThomasWaldmann> it doesn't matter whether you process huge amounts of data coming from page revisions or from attachments
81 2010-04-02T19:48:58 * ThomasWaldmann runs py.test to see if we have a failing search test
82 2010-04-02T19:49:05 <dreimark> now i am confused
83 2010-04-02T19:49:10 <ThomasWaldmann> if not, there should be one that identifies that problem
84 2010-04-02T19:50:11 * dreimark has only the test failing which esyr has not fixed
85 2010-04-02T19:50:46 <dreimark> and a pep8 failure
86 2010-04-02T19:52:15 *** valeuf
87 2010-04-02T19:52:31 * ThomasWaldmann fixes some failures
88 2010-04-02T19:54:28 <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5640:3cb7a64eaef5 1.9/MoinMoin/formatter/text_docbook.py: fix pep8 failures
89 2010-04-02T19:54:45 *** valeuf
90 2010-04-02T20:04:04 <ThomasWaldmann> 2 parser/formatter tests are broken
91 2010-04-02T20:06:03 <ThomasWaldmann> but all search tests work
92 2010-04-02T20:07:47 <dreimark> they work because we don't have much content search tests
93 2010-04-02T20:09:43 <ThomasWaldmann> can you add a simple test that fails for that bug?
94 2010-04-02T20:10:05 <dreimark> yes
95 2010-04-02T20:32:11 *** valeuf
96 2010-04-02T20:33:12 *** grzywacz
97 2010-04-02T21:13:46 *** valeuf
98 2010-04-02T21:19:17 <CIA-55> Reimar Bauer <rb.proj AT googlemail DOT com> default * 5641:a0fdc0e059cc 1.9/MoinMoin/search/_tests/test_search.py: test_search: xapian search test for regex content search
99 2010-04-02T21:22:13 <dreimark> ThomasWaldmann: a l33t string for regex search should not give any collision, the others tests i have had before currently have that problem
100 2010-04-02T21:24:11 <dreimark> i used a new page because of too much test dependencies of the crap tests
101 2010-04-02T21:52:55 * ThomasWaldmann simplifies it
102 2010-04-02T21:54:58 * ThomasWaldmann curses search tests
103 2010-04-02T22:06:30 <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5642:226e1b0aa79c 1.9/MoinMoin/search/_tests/test_search.py: search tests: simplify and extend test for regex search
104 2010-04-02T22:06:36 <ThomasWaldmann> dreimark: look there
105 2010-04-02T22:08:32 <dreimark> hmm, you killed the l33t text
106 2010-04-02T22:09:28 <dreimark> me was thinking about converting all of that strings to leet
107 2010-04-02T22:09:51 <dreimark> because after some time my testwiki is almost all a dictionary
108 2010-04-02T22:10:00 <dreimark> and i have lots of false hits
109 2010-04-02T22:11:06 <dreimark> assert [u'MoinMoinBugs/1.6.0XapianWeigthAndSortingOfRe', u'ContentSearchLower'] == ['ContentSearchLower']
110 2010-04-02T22:11:24 <dreimark> ThomasWaldmann: 4 failed, 154 passed, 1 skipped in 30.69 seconds
111 2010-04-02T22:26:33 <JosefMeier> Is it necessary to call "getPage" in a multicall session in xmlrpc? Why??? In my c# xmlrpc-lib there is no multicall support?
112 2010-04-02T22:27:18 <dreimark> yes
113 2010-04-02T22:27:32 <dreimark> because of security
114 2010-04-02T22:29:15 <JosefMeier> I used "applyAuthToken" before getPage but in no multicall session. It fails. I assume that is normal behaviour than?
115 2010-04-02T22:30:00 * ThomasWaldmann kills that len() crap in search tests
116 2010-04-02T22:30:58 <ThomasWaldmann> JosefMeier: applyAuthToken only works for same call, thus you need multicall
117 2010-04-02T22:34:17 <JosefMeier> Is there an other way for auth other than applyauthtoken which doesn't need multicalls?
118 2010-04-02T22:36:21 * dreimark the len may be not the problem besides for extending test, the words make more problems
119 2010-04-02T22:40:03 <ThomasWaldmann> no
120 2010-04-02T22:40:24 <ThomasWaldmann> len() is a major problem
121 2010-04-02T22:41:21 <ThomasWaldmann> if you search for 2 apples and i give you 2 bananas, you are satisfied?
122 2010-04-02T22:41:29 *** valeuf
123 2010-04-02T22:42:33 <dreimark> ThomasWaldmann: hehe, depends on I don't like apples
124 2010-04-02T22:43:14 <dreimark> but the search probably won't know and i get 2 apples and 2 bananas
125 2010-04-02T22:44:34 *** valeuf
126 2010-04-02T22:58:44 *** valeuf
127 2010-04-02T22:59:43 *** JosefMeier
128 2010-04-02T22:59:58 *** valeuf
129 2010-04-02T23:10:13 *** grzywacz
130 2010-04-02T23:14:37 *** JosefMeier
131 2010-04-02T23:41:26 <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5643:6282bd44c14c 1.9/MoinMoin/search/_tests/test_search.py: search tests: get rid of most crappy len() based tests
132 2010-04-02T23:59:37 <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5644:3100da24a7fa 1.9/MoinMoin/search/_tests/test_search.py: search tests: some cosmetic fixes, improve comments/docstrings
MoinMoin: MoinMoinChat/Logs/moin-dev/2010-04-02 (last edited 2010-04-01 22:15:02 by IrcLogImporter)