1 2011-07-25T00:03:52 <dreimark> gn
2 2011-07-25T00:05:12 *** sinha
3 2011-07-25T05:25:56 *** Marchael
4 2011-07-25T05:27:41 <Marchael> moin
5 2011-07-25T08:04:12 *** pkumar
6 2011-07-25T08:53:50 <dreimark> moin
7 2011-07-25T09:34:29 <pkumar> I don't understand this: unlike MemoryBackend, why for AclWrapperBackend, a new item is created for the backend even without a revision?
8 2011-07-25T09:39:28 <pkumar> can anyone please look into test_backends.test_new_item_create_revision() and help me with that?
9 2011-07-25T09:42:38 *** gwork
10 2011-07-25T09:47:16 <ThomasWaldmann> moin
11 2011-07-25T09:47:26 <ThomasWaldmann> Marchael: see EP + bug tracker
12 2011-07-25T09:48:53 <Marchael> hi ThomasWaldmann
13 2011-07-25T09:48:55 <Marchael> ok
14 2011-07-25T09:54:01 <ThomasWaldmann> and i would like to see a fix for that invisible result item name soon
15 2011-07-25T09:58:57 <Marchael> ThomasWaldmann: what causing that? iirc you told me about a patch
16 2011-07-25T10:02:38 <ThomasWaldmann> i guess your highlighting is malfunctioning, so just stop highlighting the item name until that works better
17 2011-07-25T10:03:51 <ThomasWaldmann> you could temporarily replace it by <itemlink> hl: <highlighted name>
18 2011-07-25T10:13:06 <Marchael> ThomasWaldmann: could you give me that patch, so I could see invisible names?
19 2011-07-25T10:15:09 <Marchael> I forgot remove fragmenter=formatter in search_results.html, so may be that can causing search malfunction
20 2011-07-25T10:15:26 <Marchael> s/search/highlight/
21 2011-07-25T10:16:57 <ThomasWaldmann> i have invisible names all the time when testing the stuff
22 2011-07-25T10:17:46 <ThomasWaldmann> search for language:en for example
23 2011-07-25T10:18:25 <ThomasWaldmann> you get a completely useless list of nothings
24 2011-07-25T10:18:40 <Marchael> hm, searching for now working only for "name" field
25 2011-07-25T10:19:09 <ThomasWaldmann> no, it actually works quite good and for all metadata, you just don't see the result correctly
26 2011-07-25T10:19:19 <ThomasWaldmann> (10:02) < ThomasWal> you could temporarily replace it by <itemlink> hl: <highlighted name>
27 2011-07-25T10:19:26 <Marchael> I know
28 2011-07-25T10:19:26 <ThomasWaldmann> so for now, just do it like that
29 2011-07-25T10:19:45 <Marchael> intresting
30 2011-07-25T10:20:20 <Marchael> I'm using
31 2011-07-25T10:20:20 <Marchael> > qp = QueryParser("name", schema=latest_index.schema)
32 2011-07-25T10:20:38 <Marchael> so how you could search in another fields?
33 2011-07-25T10:20:54 <ThomasWaldmann> (10:16) < ThomasWal> search for language:en for example
34 2011-07-25T10:21:20 <ThomasWaldmann> that "name" is just giving the default if you don't specify the field name
35 2011-07-25T10:22:02 <Marchael> ah
36 2011-07-25T10:22:27 * Marchael must pay more attention to reading dos
37 2011-07-25T10:22:42 <Marchael> *docs
38 2011-07-25T10:29:42 *** greg_f
39 2011-07-25T10:54:50 <Marchael> ThomasWaldmann: I just added <b></b>
40 2011-07-25T10:55:00 <Marchael> or is it wrong?
41 2011-07-25T10:58:15 <Marchael> ThomasWaldmann: I added extended search form as you asked http://i.imgur.com/9OtAC.png
42 2011-07-25T11:01:45 *** pkumar
43 2011-07-25T11:03:16 <Marchael> http://codereview.appspot.com/4798052
44 2011-07-25T11:03:37 <Marchael> ThomasWaldmann: please, review ^
45 2011-07-25T11:18:53 * dreimark added some comments
46 2011-07-25T11:21:33 <Marchael> dreimark: thx
47 2011-07-25T12:09:31 * ThomasWaldmann reviews
48 2011-07-25T12:19:03 <ThomasWaldmann> Marchael: review done
49 2011-07-25T12:19:24 * ThomasWaldmann must go, bbl
50 2011-07-25T12:51:39 *** Marchael
51 2011-07-25T12:56:51 *** Marchael
52 2011-07-25T13:55:16 <Marchael> ThomasWaldmann: http://codereview.appspot.com/4798052/ patchset 2
53 2011-07-25T15:17:35 <ThomasWaldmann> Marchael: mostly ok, one naming thing though
54 2011-07-25T15:59:50 <Marchael> > start_page=results.pagenum, end_page=results.pagecount
55 2011-07-25T15:59:50 <Marchael> ThomasWaldmann: you mean use pagenum=..., pagecount=...?
56 2011-07-25T16:01:30 <Marchael> if so , may I commit?
57 2011-07-25T16:04:11 <ThomasWaldmann> yes I mean that
58 2011-07-25T16:04:17 <Marchael> ok
59 2011-07-25T16:04:18 <ThomasWaldmann> yes
60 2011-07-25T16:06:32 <Marchael> ok, I test it now before push
61 2011-07-25T16:14:58 <Marchael> seems work
62 2011-07-25T16:15:09 <Marchael> at least moin runs :)
63 2011-07-25T16:15:27 <CIA-114> Michael Mayorov <marchael@kb.csu.ru> * fb400cd3391f r354 moin-2.0/MoinMoin/ (3 files in 3 dirs): Add long search input field on search results page, remove highlighing search results due bugs, some fixes in css
64 2011-07-25T16:21:20 *** brunomartin
65 2011-07-25T16:31:26 *** brunomartin
66 2011-07-25T16:31:46 *** brunomartin
67 2011-07-25T16:34:24 *** brunomartin
68 2011-07-25T16:34:31 *** brunomartin
69 2011-07-25T16:40:16 *** pkumar
70 2011-07-25T16:51:51 <pkumar> xorAxAx : hi, I think that the problem with router backend tests is indexing.IndexingBackendMixin.create_item. If we have item.publish_metadata() self.backend.has_item(u'somevalue') becomes true
71 2011-07-25T16:52:07 <pkumar> xorAxAx : there's also a test for this
72 2011-07-25T16:52:38 <xorAxAx> what exactly does the indexing mixin do that confuses the fs backend?
73 2011-07-25T16:55:14 <pkumar> xorAxAx : in the IndexingBackendMixin.create_item() we add UUID and name to item, and publish it. Now, this ends up in self.backend.has_item(u'somevalue') = True, which was supposed to be False as per the test
74 2011-07-25T16:55:39 <xorAxAx> huh, create_item publishes the item?!
75 2011-07-25T16:56:22 <pkumar> yeah, so we get an item in the backend but he test suggest it should not.
76 2011-07-25T16:57:07 <xorAxAx> that sounds utterly broken
77 2011-07-25T16:57:28 <pkumar> though, for fs2 when we use only FS2Bakend for create_backend we don't have all those errors and failures
78 2011-07-25T16:57:55 <pkumar> but for router backend we have that as this uses index mixin
79 2011-07-25T16:57:58 <pkumar> I guess
80 2011-07-25T16:58:15 <xorAxAx> why does the router backend require it?
81 2011-07-25T16:59:21 <pkumar> to add UUID and itemname I guess as the code suggest.
82 2011-07-25T16:59:42 <pkumar> also, I'm not able to quite follow that part of execution flow
83 2011-07-25T17:01:12 <pkumar> what exactly is calling IndexingBAckendMixin.create_item()
84 2011-07-25T17:01:21 <pkumar> ?
85 2011-07-25T17:07:20 <ThomasWaldmann> Marchael: did you move the small search input field in the theme? it seems to be a bit more left now.
86 2011-07-25T17:08:50 <ThomasWaldmann> Marchael: remove the artefact (that appears above the big search input line)
87 2011-07-25T17:10:52 <ThomasWaldmann> Marchael: if we are searching only in the current revisions (== as now), maybe better link to the item url without revno. if you link using the revision number, it'll show that rev navigation at the top and that is not needed in this case usually.
88 2011-07-25T17:11:59 <ThomasWaldmann> contenttype:text/x-rst search yields unexpected results
89 2011-07-25T17:12:38 <ThomasWaldmann> contenttype:text crashes
90 2011-07-25T17:13:46 <ThomasWaldmann> Marchael: give the itemname to the template to replace that "handlers need to give it"
91 2011-07-25T17:14:00 <Marchael> 1. hm, looks so.
92 2011-07-25T17:14:15 <Marchael> 2. it's not artefact, it's a search tip :)
93 2011-07-25T17:14:34 <Marchael> 3. ok
94 2011-07-25T17:15:02 <ThomasWaldmann> i don't think we want to permanently show a search tip there.
95 2011-07-25T17:15:14 <Marchael> contenttype:text/x-rst working normal
96 2011-07-25T17:15:26 <ThomasWaldmann> no, it finds all text items, not just rst
97 2011-07-25T17:15:38 <Marchael> ah
98 2011-07-25T17:16:22 <ThomasWaldmann> you can fix that crash and that malfunctioning in one go. just first think about what needs to be done.
99 2011-07-25T17:17:21 <Marchael> Input: text/x.moin.wiki;charset=utf-8Output: "text", "x.moin.wiki", "charset=utf-8"
100 2011-07-25T17:17:27 <Marchael> Input: text/x.moin.wiki;charset=utf-8
101 2011-07-25T17:17:27 <Marchael> Output: "text", "x.moin.wiki", "charset=utf-8"
102 2011-07-25T17:17:45 <Marchael> so that's why searcher found all docs
103 2011-07-25T17:18:54 <Marchael> ThomasWaldmann: so what's wrong?
104 2011-07-25T17:21:02 <ThomasWaldmann> well, it is explainable, but still not the behaviour a user would expect
105 2011-07-25T17:21:10 <Marchael> hm
106 2011-07-25T17:21:23 <Marchael> then I'm need to make some changes to analyzer =\
107 2011-07-25T17:21:53 <ThomasWaldmann> maybe we need also index and query mode
108 2011-07-25T17:22:13 <Marchael> ThomasWaldmann: may use ID for contenttype?
109 2011-07-25T17:22:47 <ThomasWaldmann> for index, it should tokenize and also yield full mimetype and full contenttype
110 2011-07-25T17:23:07 <ThomasWaldmann> for searching, it should use as is
111 2011-07-25T17:23:31 <Marchael> ok, then we need MultiFiler
112 2011-07-25T17:23:57 <ThomasWaldmann> so that a user can search for contenttype:text and will find text/*
113 2011-07-25T17:24:30 <ThomasWaldmann> but if he searches for text/x-rst, he of course only wants text/x-rst, not text/*
114 2011-07-25T17:24:32 <Marchael> hm, but searching still fails for wildcards
115 2011-07-25T17:26:53 <ThomasWaldmann> hmm, let me look at tokenizer...
116 2011-07-25T17:27:24 <xorAxAx> pkumar: find out who wrote it (via annotate) and ask the author :)
117 2011-07-25T17:28:18 <pkumar> xorAxAx : I'm sorry but what's annotate?
118 2011-07-25T17:28:22 <ThomasWaldmann> didn't I already tell not to waste time on itemlevel metadata?
119 2011-07-25T17:28:53 <xorAxAx> pkumar: a DVCS feature ... just click annotate on bitbucket and you will see who wrote every single line
120 2011-07-25T17:29:52 <ThomasWaldmann> Marchael: i think the tokenizer could yield type, type/subtype and then params
121 2011-07-25T17:30:05 <ThomasWaldmann> for indexing
122 2011-07-25T17:30:15 <pkumar> ThomasWaldmann : yeah, but that that one fix could fix almost all the failures of storage, so I thought ..., also the task for the week was for fixing stuffs in storage.
123 2011-07-25T17:30:20 <ThomasWaldmann> and for query, just use "as is"
124 2011-07-25T17:30:51 *** gwork
125 2011-07-25T17:30:56 <Marchael> ThomasWaldmann: hm, what poinrt of it? user could user text/* for searching all text docs
126 2011-07-25T17:31:02 <Marchael> *point
127 2011-07-25T17:31:35 <Marchael> *user could use
128 2011-07-25T17:31:54 <pkumar> ThomasWaldmann : will schedule new task :)
129 2011-07-25T17:31:54 <pkumar> xorAxAx : thanks,
130 2011-07-25T17:32:30 <Marchael> or rather just text*
131 2011-07-25T17:32:39 <Marchael> contenttype:text*
132 2011-07-25T17:34:45 <Marchael> ThomasWaldmann: ^
133 2011-07-25T17:35:45 <Marchael> so tokenizer just yield type/subtype, yield params
134 2011-07-25T17:43:53 <ThomasWaldmann> Marchael: yes, one can always use wildcard search, but that's not very efficient
135 2011-07-25T17:44:55 <ThomasWaldmann> thus i think we want to be able to find the major types directly (or even some specific groups of mimetypes, but that can be done later)
136 2011-07-25T17:46:03 <Marchael> ok
137 2011-07-25T17:53:53 <ThomasWaldmann> hmm, somehow i can't make sense of contenttype vs. name behaviour yet
138 2011-07-25T17:54:50 <ThomasWaldmann> e.g. if you have item names FooBar and BarBaz and you search for FooBar, it'll find FooBar only (although it tokenized that to "Foo", "Bar"), so it looks like it needed both tokens to be found.
139 2011-07-25T17:55:51 <ThomasWaldmann> but if you search for contenttype:text/x.moin.wiki, it'll find all text stuff, so it seems to be happy with finding just one of the pieces.
140 2011-07-25T17:55:56 <Marchael> may be it search for "Foo" AND "Bar"
141 2011-07-25T17:56:29 <ThomasWaldmann> well, why not text AND x.moin.wiki then?
142 2011-07-25T17:57:13 <ThomasWaldmann> i feel we need first to be able to explain that :)
143 2011-07-25T17:59:37 <pkumar> bbl
144 2011-07-25T17:59:42 *** pkumar
145 2011-07-25T18:04:59 <Marchael> rather strange
146 2011-07-25T18:45:50 *** pkumar
147 2011-07-25T19:06:52 <Marchael> ok, patch was accepted. https://bitbucket.org/mchaput/whoosh/issue/156/with-patch-highlighting-fails-for
148 2011-07-25T19:17:19 <CIA-114> pkumar <contactprashantat@gmail.com> * 30f4fb350ad1 r421 pytest2/MoinMoin/storage/_tests/test_backends.py: fixed errors in test_backends_fs2 and few failures in storage module
149 2011-07-25T19:17:19 <CIA-114> Himanshu Awasthi <89hardy.code@gmail.com> * 462163be25bc r422 pytest2/MoinMoin/storage/_tests/test_backends.py: test failure removed from test_middleware_acl.py
150 2011-07-25T19:17:19 <CIA-114> Himanshu Awasthi <89hardy.code@gmail.com> * 4816f4025add r423 pytest2/MoinMoin/storage/__init__.py: Removal of errors from test_middleware_acl.py and test_backends_router.py
151 2011-07-25T19:17:19 <CIA-114> Himanshu Awasthi <89hardy.code@gmail.com> * 5279825c90b2 r424 pytest2/MoinMoin/storage/_tests/test_serialization.py: Fixed the entire test module comparing XML manually using xml.dom.minidom because xmldiff did not help
152 2011-07-25T19:17:20 <CIA-114> Himanshu Awasthi <89hardy.code@gmail.com> * fae6e626eb19 r425 pytest2/MoinMoin/storage/_tests/test_backends.py: Fixed memory backend and sqla test for creating an item and a revision. Rollback was not required, did an item commit instead.
153 2011-07-25T19:17:20 <CIA-114> Himanshu Awasthi <89hardy.code@gmail.com> * d9eb2bd90918 r426 pytest2/MoinMoin/storage/ (_tests/test_backends_router.py backends/router.py): Fixed rename method in router backend in turn fixing a failed test.
154 2011-07-25T19:17:21 <CIA-114> pkumar <contactprashantat@gmail.com> * ca82715a8bf8 r427 pytest2/MoinMoin/storage/ (5 files in 3 dirs): merged with hardy89's repo
155 2011-07-25T19:17:21 <CIA-114> pkumar <contactprashantat@gmail.com> * fca6f30612d1 r428 pytest2/MoinMoin/storage/ (_tests/test_backends.py __init__.py): corrected wrong statement from merge
156 2011-07-25T19:17:53 <CIA-114> pkumar <contactprashantat@gmail.com> * 09b38f6280ae r429 pytest2/MoinMoin/storage/_tests/ (test_backends.py test_backends_fs2.py): FS2Backend is sufficient for testing storage.backends.fs2
157 2011-07-25T19:18:54 *** sinha
158 2011-07-25T19:31:47 <ThomasWaldmann> Marchael: ah, good :)
159 2011-07-25T19:40:14 *** sinha
160 2011-07-25T19:51:07 *** brunomartin
161 2011-07-25T19:51:49 *** raignarok
162 2011-07-25T19:53:55 *** greg_f
163 2011-07-25T19:55:18 *** brunomartin
164 2011-07-25T20:00:50 <CIA-114> Thomas Waldmann <tw AT waldmann-edv DOT de> * 6ecab9748faa r430 pytest2/MoinMoin/translations/MoinMoin.pot: update MoinMoin.pot
165 2011-07-25T20:00:50 <CIA-114> Thomas Waldmann <tw AT waldmann-edv DOT de> * 46e37e4197d4 r431 pytest2/MoinMoin/translations/de/LC_MESSAGES/messages.po: updated german translation
166 2011-07-25T20:00:50 <CIA-114> Thomas Waldmann <tw AT waldmann-edv DOT de> * ebdfe9448e01 r432 pytest2/MoinMoin/ (22 files in 3 dirs): (log message trimmed)
167 2011-07-25T20:00:51 <CIA-114> make converter api more homogeneous, adapt tests
168 2011-07-25T20:00:51 <CIA-114> you can give a file-like object (like a revision), a str or unicode now to a
169 2011-07-25T20:00:51 <CIA-114> input converter (no matter for what mimetype it is).
170 2011-07-25T20:00:51 <CIA-114> you also need to give the contenttype (at least for the text input converters,
171 2011-07-25T20:00:51 <CIA-114> they will extract the charset from there and decode appropriately).
172 2011-07-25T20:00:52 <CIA-114> removed the feed_input_conv methods, we always give the revision and the
173 2011-07-25T20:00:52 <CIA-114> Thomas Waldmann <tw AT waldmann-edv DOT de> * e02f3f45ab5d r433 pytest2/MoinMoin/converter/ (16 files): converters: cosmetic changes
174 2011-07-25T20:00:53 <CIA-114> Thomas Waldmann <tw AT waldmann-edv DOT de> * 5158f027bb3f r434 pytest2/MoinMoin/ (converter/__init__.py apps/frontend/views.py): (log message trimmed)
175 2011-07-25T20:00:53 <CIA-114> use converters to convert a rev to indexable content
176 2011-07-25T20:00:54 <CIA-114> revision data may contain all sorts of content-types, e.g.:
177 2011-07-25T20:00:55 <CIA-114> - markup text (moin, creole, docbook, rst, html, ...)
178 2011-07-25T20:00:55 <CIA-114> - plain text (code, docs, ...)
179 2011-07-25T20:00:56 <CIA-114> - images, audio
180 2011-07-25T20:02:50 <Marchael> great commit!
181 2011-07-25T20:04:02 *** sinha
182 2011-07-25T20:28:35 <pkumar> bbl
183 2011-07-25T20:33:53 *** sinha
184 2011-07-25T20:41:53 *** pkumar
185 2011-07-25T20:52:16 *** sinha
186 2011-07-25T20:54:02 <ThomasWaldmann> Marchael: that was just a merge of some old changesets into pytest2 repo
187 2011-07-25T20:54:46 * Marchael migrates tokenizer to filter
188 2011-07-25T20:54:57 <Marchael> *converts
189 2011-07-25T20:55:08 <Marchael> *converting
190 2011-07-25T20:55:10 <Marchael> ah
191 2011-07-25T21:05:42 <CIA-114> Akash Sinha <akash2607@gmail.com> * b9f0ba2ba779 r337 default/MoinMoin/ (3 files in 3 dirs): paging code at item history moved under util, such that this function could be called from anywhere else.
192 2011-07-25T21:05:57 <Marchael> sinha: thx
193 2011-07-25T21:06:33 <sinha> Marchael: welcome, please see that history.html for the frontend things
194 2011-07-25T21:36:27 <Marchael> argh
195 2011-07-25T21:36:59 <Marchael> "contenttype:text" fails but "contenttype:text/" not
196 2011-07-25T21:37:32 <Marchael> ThomasWaldmann: I recieve traceback from mime.py
197 2011-07-25T21:38:19 <Marchael> but use unlikely will enter one '/' in the end
198 2011-07-25T21:38:23 <Marchael> *user
199 2011-07-25T21:38:59 <Marchael> ThomasWaldmann: may be need some changes in mime.py?
200 2011-07-25T21:39:53 <Marchael> now "contenttype:text/x-rst" gives expected results
201 2011-07-25T21:41:22 <Marchael> or I should check this '/' in analyzer?
202 2011-07-25T21:48:07 <Marchael> seems like I found solution
203 2011-07-25T21:55:22 <Marchael> ThomasWaldmann: done
204 2011-07-25T21:56:34 <Marchael> but it's still fails for *
205 2011-07-25T22:01:52 <ThomasWaldmann> Marchael: show me the code you have
206 2011-07-25T22:02:09 <Marchael> ThomasWaldmann: http://codereview.appspot.com/4822050
207 2011-07-25T22:03:41 <ThomasWaldmann> why is it called filter now?
208 2011-07-25T22:04:20 <Marchael> because it's not tokenizer
209 2011-07-25T22:04:34 <Marchael> Type class is
210 2011-07-25T22:04:50 <Marchael> this class just filtering that output
211 2011-07-25T22:05:09 <Marchael> this = ContentTypeFilter
212 2011-07-25T22:21:07 <ThomasWaldmann> done
213 2011-07-25T22:27:40 <xorAxAx> waldi: i asked you something
214 2011-07-25T22:32:29 <Marchael> ThomasWaldmann: answered
215 2011-07-25T22:40:30 <ThomasWaldmann> maybe if you think about the ??? stuff, you'll find some issues
216 2011-07-25T22:42:33 <ThomasWaldmann> and I am still missing a plan, docs, etc.
217 2011-07-25T22:47:37 <Marchael> ThomasWaldmann: ep 41
218 2011-07-25T22:52:40 <ThomasWaldmann> for having same output as input, you use quite a lot of code
219 2011-07-25T22:53:41 <Marchael> you think what I can just pass "query" as is, without parsing?
220 2011-07-25T22:55:04 <ThomasWaldmann> well, you have to have a plan what you really want to do. i didn't see one yet.
221 2011-07-25T22:55:28 <ThomasWaldmann> and you should also know what your current problem is before making that plan.
222 2011-07-25T23:01:40 <ThomasWaldmann> brb
223 2011-07-25T23:04:13 <Marchael> ThomasWaldmann: ep 62
224 2011-07-25T23:17:07 <ThomasWaldmann> why does search ignore subtype?
225 2011-07-25T23:17:27 <Marchael> *ignored
226 2011-07-25T23:17:57 <dreimark> re
227 2011-07-25T23:18:52 <Marchael> because document with different subtypes having one type
228 2011-07-25T23:19:01 <Marchael> *documents
229 2011-07-25T23:19:30 <Marchael> hm
230 2011-07-25T23:19:47 <Marchael> seems like whoosh using OR in that case
231 2011-07-25T23:20:30 <Marchael> in other case, like with "name" it uses AND when parsing tokens
232 2011-07-25T23:21:42 <Marchael> may be I just should queryparser parameter to use AND
233 2011-07-25T23:21:52 <Marchael> *should use
234 2011-07-25T23:25:04 <Marchael> ThomasWaldmann: ^
235 2011-07-25T23:28:18 <ThomasWaldmann> that's just guessing, you can't point at the problem yet
236 2011-07-25T23:37:14 * Marchael sleeping
237 2011-07-25T23:37:53 <Marchael> need to sleep few hours, then may be I could uderstand what may cause it.
238 2011-07-25T23:37:59 <Marchael> gn/bbl
239 2011-07-25T23:47:17 <dreimark> gn Marchael
240 2011-07-25T23:50:27 *** sinha
241 2011-07-25T23:51:44 <dreimark> ThomasWaldmann: summit mails from carol
