1 2010-05-24T00:08:21 <dreimark> diofeher: can your irc client stay online?
2 2010-05-24T00:09:21 <diofeher> dreimark: all day?
3 2010-05-24T00:09:30 <dreimark> yes
4 2010-05-24T00:09:51 <diofeher> no.. but my friends that entered in gsoc too will install a psybnc to me
5 2010-05-24T00:09:52 <dreimark> you can use irssi with screen
6 2010-05-24T00:10:32 <diofeher> mm, i use chatzilla here
7 2010-05-24T00:10:58 <dreimark> if you have access to a system which is always online you could run irssi in screen
8 2010-05-24T00:11:05 <ThomasWaldmann> diofeher: if that works out with psybnc, that's fine
9 2010-05-24T00:11:45 <ThomasWaldmann> if you have a flat rate, you should be connected as permanently as possible
10 2010-05-24T00:12:41 <dreimark> the benifit of that is that everyone can leave messages and noone has to wait until all are online
11 2010-05-24T00:13:03 <diofeher> ok! i will install psybnc as soon as possible
12 2010-05-24T00:13:12 <diofeher> until wednesday i think
13 2010-05-24T00:19:08 <ThomasWaldmann> what dreimark meant is running irsii in screen via ssh on a permanently connected system
14 2010-05-24T00:21:28 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8225:9406c50a2a53 2.0-dev/MoinMoin/storage/serialization.py: storage serialization: use py3-like element names: str and bytes
15 2010-05-24T00:21:31 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8226:8d18b007dbb0 2.0-dev/ (MoinMoin/_tests/testitems.xml wiki/syspages.xml): system pages / test pages: update xml files
16 2010-05-24T00:25:20 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8227:19c91ee08ff2 2.0-dev/MoinMoin/storage/_tests/test_serialization.py: storage serialization: adapt tests to py3-like element names
17 2010-05-24T00:57:10 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8228:d70bfb384a12 2.0-dev/MoinMoin/items/__init__.py: more metadata type fixes
18 2010-05-24T00:58:47 <valeuf> xorAxAx: just that order of the tag does not always matter
19 2010-05-24T00:59:35 <valeuf> xorAxAx: it is pretty obscure for me at this time, but I did not check that correctly
20 2010-05-24T00:59:51 <valeuf> xorAxAx: I will do it more seriously tomorrow
21 2010-05-24T01:00:16 <xorAxAx> valeuf: i dont understand the order thing
22 2010-05-24T01:04:26 <valeuf> <em><strong>My Bold and Italic text</strong></em> is equivalent to <strong><em>My Bold and Italic Text</strong></em>
23 2010-05-24T01:04:48 <valeuf> a good test should be ok for the two situation because it does not matter
24 2010-05-24T01:05:06 <valeuf> but at this time I do not see how to do it
25 2010-05-24T01:05:23 <valeuf> otherwise, I think it should be ok, I just need to code it
26 2010-05-24T01:10:43 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8229:9a2038815645 2.0-dev/MoinMoin/support/kvstore.py: MoinMoin.support.kvstore - a flexible key/value store implemented with sqlalchemy
27 2010-05-24T01:14:09 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8230:48ecb33067cd 2.0-dev/MoinMoin/storage/backends/indexing.py: storage: add indexing mixin classes
28 2010-05-24T01:14:35 <diofeher> valeuf: maybe you can use selenium...
29 2010-05-24T01:15:21 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8231:810932eeffcd 2.0-dev/MoinMoin/storage/backends/ (__init__.py fs2.py): use indexing mixins for fs2 backend (others might be slightly broken)
30 2010-05-24T01:15:28 <diofeher> valeuf: i think it will be useful for you, http://seleniumhq.org/docs/
31 2010-05-24T01:17:16 <valeuf> diofeher: it can be a solution, but I am sure it is possible to do it with Xpath
32 2010-05-24T01:18:56 <diofeher> valeuf: you can go through a Html document with xpath? or have to convert to xml?
33 2010-05-24T01:19:26 <xorAxAx> valeuf: well, i think it is indeed not possible. but if your code maintains the order, it should be fine
34 2010-05-24T01:23:04 <ThomasWaldmann> btw, metadata indexing stuff basically works now
35 2010-05-24T01:23:52 <ThomasWaldmann> (for fs2 backend, please review)
36 2010-05-24T01:24:54 <ThomasWaldmann> some pieces are missing still, e.g. item uuid/name, using the index, ...
37 2010-05-24T01:37:02 <valeuf> diofeher: well, we output Xhtml, so yes normally
38 2010-05-24T01:37:26 <valeuf> xorAxAx: yep, anyway I thought that I won't care about the order at this time
39 2010-05-24T01:39:45 <diofeher> ah, nice :]
40 2010-05-24T02:37:18 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8232:44b2e558e4e3 2.0-dev/MoinMoin/ (4 files in 2 dirs): add in-memory indexing to MemoryBackend, fix some tests
41 2010-05-24T02:51:00 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8233:8e7836f5fd13 2.0-dev/MoinMoin/storage/backends/fs.py: fs backend: add indexing mixins
42 2010-05-24T03:06:03 <valeuf> ThomasWaldmann: you are a night bird
43 2010-05-24T04:08:14 *** ei-grad
44 2010-05-24T04:21:30 *** DmitryAndreev
45 2010-05-24T06:01:46 *** diofeher
46 2010-05-24T06:30:34 *** diofeher
47 2010-05-24T07:09:07 <diofeher> i made some changes in http://www.moinmo.in/Jinja 2010
48 2010-05-24T07:09:20 <diofeher> if anyone can answer later, it will be very helpful :)
49 2010-05-24T07:15:15 <diofeher> going to sleep... be back in 6 hours
50 2010-05-24T07:15:25 *** diofeher
51 2010-05-24T09:18:54 <eSyr> moin
52 2010-05-24T09:28:02 *** kursor
53 2010-05-24T09:35:04 *** DmitryAndreev
54 2010-05-24T09:55:17 <dreimark> moin
55 2010-05-24T09:55:22 <dreimark> ThomasWaldmann: http://moinmo.in/FeatureRequests/XMLRPC needs no editor backup
56 2010-05-24T09:55:35 <dreimark> http://moinmo.in/FeatureRequests/XMLRPC%20needs%20no%20editor%20backup
57 2010-05-24T09:56:44 <dreimark> the memory my testsystem needs decreased from 5.5GB to 1.7GB
58 2010-05-24T09:56:51 <dreimark> s/needs//
59 2010-05-24T10:06:06 * dreimark believes that drafts have a bigger problem
60 2010-05-24T10:08:18 <dreimark> the moin server process needs less than 1.7GB (this was the value the monitor told)
61 2010-05-24T10:18:51 <dreimark> (the wiki hat > 60K pages)
62 2010-05-24T10:18:59 <dreimark> s/hat/has/
63 2010-05-24T10:43:50 <CIA-47> Reimar Bauer <rb.proj AT googlemail DOT com> default * 5672:203afc609f8a 1.9/MoinMoin/xmlrpc/__init__.py:
64 2010-05-24T10:43:50 <CIA-47> xmlrpc: Disable editor backup for putPage, renamePage and revertPage because if
65 2010-05-24T10:43:50 <CIA-47> pages become uploaded by xmlrpc then the draft cache arena can fast increase to
66 2010-05-24T10:43:50 <CIA-47> a big cache file size. On each of these methods called and on each server
67 2010-05-24T10:43:50 <CIA-47> process the cache file must be loaded and processed. This can costs much memory
68 2010-05-24T10:43:51 <CIA-47> and decreases speed of these calls.
69 2010-05-24T10:47:53 * dreimark thinks we should limit the draft arena to a few MBytes
70 2010-05-24T11:58:22 <ThomasWaldmann> moin
71 2010-05-24T11:58:49 <ThomasWaldmann> dreimark: what's the issue with drafts?
72 2010-05-24T12:02:27 <ThomasWaldmann> (draft is expected to hold one copy of an edited page, why is it so big for you?)
73 2010-05-24T12:04:56 *** kursor
74 2010-05-24T12:07:30 <ThomasWaldmann> aa_: (07:09) < diofeher> i made some changes in http://www.moinmo.in/Jinja 2010
75 2010-05-24T12:07:42 <ThomasWaldmann> maybe you can answer that best
76 2010-05-24T12:20:13 <aa_> ThomasWaldmann: looking now
77 2010-05-24T12:27:07 <ThomasWaldmann> BTW, SOC coding phase has officially begun today.
78 2010-05-24T12:27:46 <xorAxAx> yes, where are the commits? :)
79 2010-05-24T12:27:52 <TheSheep> ha
80 2010-05-24T12:27:57 <TheSheep> where are the students?
81 2010-05-24T12:28:01 <DmitryAndreev> here
82 2010-05-24T12:28:05 <DmitryAndreev> .)
83 2010-05-24T12:28:07 <TheSheep> \o/
84 2010-05-24T12:29:38 <ThomasWaldmann> still some hours left for first code commit or wiki calendar log page :)
85 2010-05-24T12:32:40 <DmitryAndreev> ow, now i'm think how to organize wiki_out convertor. html_out looks to heavy.
86 2010-05-24T12:34:08 <DmitryAndreev> also esyr want's from me some review of parsing systems with code exempls
87 2010-05-24T12:35:05 <waldi> DmitryAndreev: you need to visit every element within a tree. there are two approches: DFS and BFS. only the first is really usable in this case and this is what the html converter does
88 2010-05-24T12:36:20 <ThomasWaldmann> btw, did anyone work on a image converter?
89 2010-05-24T12:40:51 <DmitryAndreev> waldi, still, so many functions in html_out.) i will start without looking on attributes. attributes controller will be later.
90 2010-05-24T12:41:55 <DmitryAndreev> sorry, is my english understandable?)
91 2010-05-24T12:42:11 <TheSheep> it's very good
92 2010-05-24T12:42:14 <TheSheep> keep it up
93 2010-05-24T12:48:39 <ThomasWaldmann> btw, for daily english training I use slashdot.org, planet.python.org, bbc news and lots of stuff found by google. video.google.com "google techtalks" is nice for spoken english (mostly).
94 2010-05-24T12:49:49 <ThomasWaldmann> ah, and irc :D
95 2010-05-24T13:03:27 <TheSheep> although irc gives you a strong, recognizable accent ;)
96 2010-05-24T13:04:32 <zenhase> you got to admit, that there are several different dialects roaming the webs ;)
97 2010-05-24T13:04:52 <zenhase> i mean there is nerdspeak and lolspeak for example ;)
98 2010-05-24T13:05:42 <zenhase> and youtube-comment-speak, which must be some equivalent of redneck american slang in the virtual world
99 2010-05-24T13:08:15 <dreimark> ThomasWaldmann: because xmlrpc creates or updates many pages
100 2010-05-24T13:08:57 <dreimark> and if i run many processes i get it all the time processed
101 2010-05-24T13:09:20 <dreimark> this costs much memory
102 2010-05-24T13:37:00 <ThomasWaldmann> dreimark: so how does this mage gigabytes of RAMß
103 2010-05-24T13:37:02 <ThomasWaldmann> ?
104 2010-05-24T13:37:10 <ThomasWaldmann> *make
105 2010-05-24T13:37:25 <ThomasWaldmann> (or disk storage?)
106 2010-05-24T13:37:38 <ThomasWaldmann> please clarify
107 2010-05-24T13:38:05 <dreimark> it becomes open and read into RAM on any apache process
108 2010-05-24T13:39:26 <dreimark> and on write it becomes duplicated
109 2010-05-24T13:40:40 <dreimark> the disk storage is explainable because the user has modified 65K pages
110 2010-05-24T13:41:48 <dreimark> but I don't know exactly currently why all my processes together with that 140MB file gives in total > 5GB for the apache processes
111 2010-05-24T13:45:39 <ThomasWaldmann> ok, so it is 1 user 65K pages edited * ~2KB per page draft = 140MB user draft cache
112 2010-05-24T13:46:33 <ThomasWaldmann> please check where it loads the draft cache and whether it can be garbage collected
113 2010-05-24T13:47:18 <dreimark> yes, we need that for all other caches too
114 2010-05-24T13:56:43 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8234:0bc854eeef9a 2.0-dev/MoinMoin/storage/backends/fs.py: fs backend: use r/w file open mode for files we keep open
115 2010-05-24T13:56:56 <ThomasWaldmann> there is some chance that fs works now
116 2010-05-24T14:00:25 <aa_> ThomasWaldmann: its an empty wiki page about jinja, should I write some stuff in there?
117 2010-05-24T14:03:28 <ThomasWaldmann> aa_: i meant the questions aboubt wtforms, see info action / recent changes
118 2010-05-24T14:14:24 * ThomasWaldmann tries to fix "ProgrammingError: (ProgrammingError) SQLite objects created in a thread can only be used in that same thread." now
119 2010-05-24T14:16:15 *** ThomasWaldmann
120 2010-05-24T14:18:13 *** ThomasWaldmann
121 2010-05-24T15:24:41 <ThomasWaldmann> hmm, implicit execution is easiest for fixing
122 2010-05-24T15:26:35 <TheSheep> ThomasWaldmann: I have it fixed in hatta
123 2010-05-24T15:28:07 <TheSheep> ThomasWaldmann: http://hg.hatta-wiki.org/hatta-dev/file/09bf3c67d528/hatta.py#l1352
124 2010-05-24T15:30:11 <ThomasWaldmann> sqla does connection pooling anyway
125 2010-05-24T15:39:23 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8235:ea821ee0349c 2.0-dev/MoinMoin/ (storage/backends/indexing.py support/kvstore.py):
126 2010-05-24T15:39:23 <CIA-47> indexing storage backend mixin, kvstore: use implicit execution (see below)
127 2010-05-24T15:39:23 <CIA-47> sqlalchemy's implicit execution implicitly creates a connection to execute
128 2010-05-24T15:39:23 <CIA-47> the sql. as sqlalchemy uses a pool of connections for this (and does not
129 2010-05-24T15:39:23 <CIA-47> really create a new database connection each time), this seems to be an easy
130 2010-05-24T15:39:24 <CIA-47> way to avoid threading issues (a connection may only be used by 1 thread)
131 2010-05-24T15:39:25 <CIA-47> and also makes shorter code.
132 2010-05-24T15:40:41 <ThomasWaldmann> hmm, somehow preload_xml currently always preloads the backend storage
133 2010-05-24T15:44:09 <ThomasWaldmann> ehrm, no, it is likely because MemoryBackend also maintains an index now (which might be slightly superfluous in that case)
134 2010-05-24T15:47:20 <ThomasWaldmann> preloaded_xml = None
135 2010-05-24T15:47:46 <ThomasWaldmann> ^^ once you have loaded your favourite backend and you rather like faster startup
136 2010-05-24T16:15:30 *** ei-grad
137 2010-05-24T16:24:13 *** diofeher
138 2010-05-24T16:24:40 <diofeher> moin!
139 2010-05-24T16:28:16 <ei-grad> moin
140 2010-05-24T16:41:16 <diofeher> hey guys, do you use kdiff3 with mercurial?
141 2010-05-24T16:41:31 <diofeher> ei-grad hi :)
142 2010-05-24T16:45:36 <ei-grad> heh... on Defcon CTF quals only Chelyabinsk had 2 teams from one city, of Russian teams... 24 and 52 ranks... (my team is 52)... 4th of teams from Russia...
143 2010-05-24T16:46:32 <ThomasWaldmann> diofeher: i personally prefer conflict markers in src
144 2010-05-24T16:53:29 <diofeher> nice ThomasWaldmann..
145 2010-05-24T16:53:53 <dreimark> ei-grad: congrats - how is it going with the ldap test ?
146 2010-05-24T16:54:19 <dreimark> diofeher: me does
147 2010-05-24T16:54:30 <diofeher> well, i thinking about forms... ali ashfar answered me that he thinks the most complete project found by him is flatland
148 2010-05-24T16:54:34 <diofeher> congratz ei-grad :D
149 2010-05-24T16:55:19 <eSyr> ei-grad: huh, you're participating in CTF?
150 2010-05-24T16:55:42 <ei-grad> dreimark: i will start coding today
151 2010-05-24T16:55:50 <ei-grad> eSyr: yes)
152 2010-05-24T16:58:54 <eSyr> ei-grad: cool.
153 2010-05-24T17:04:25 <ei-grad> Team Smoked Chicken is from my university (originally, but they participate as from South Ural State University now)... They were originators of CTF in Russia, together with a team HackerDom.
154 2010-05-24T17:12:02 *** ei-grad
155 2010-05-24T17:14:27 *** ei-grad
156 2010-05-24T17:20:23 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8236:72f6db070642 2.0-dev/MoinMoin/storage/backends/ (fs.py fs2.py hg.py indexing.py memory.py sqla.py):
157 2010-05-24T17:20:23 <CIA-47> move rev.timestamp assignment from storage backends to indexing mixin
158 2010-05-24T17:20:23 <CIA-47> To index metadata, we need a valid timestamp. So if it is still None when
159 2010-05-24T17:20:23 <CIA-47> reaching the indexer, it is initialized to current time. Similar code that
160 2010-05-24T17:20:23 <CIA-47> was present in all backends was removed from there.
161 2010-05-24T17:20:24 <CIA-47> Note: breaks backends without indexing mixin if rev.timestamp is None.
162 2010-05-24T17:27:53 *** DmitryAndreev
163 2010-05-24T17:41:39 <diofeher> going to college, bye!
164 2010-05-24T17:41:42 *** diofeher
165 2010-05-24T18:14:27 <TheSheep> missed again
166 2010-05-24T18:15:32 *** ei-grad
167 2010-05-24T18:30:34 *** kursor
168 2010-05-24T18:52:35 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8237:e60ffe82eb8a 2.0-dev/MoinMoin/storage/backends/fs2.py: fs2 backend: make Item and Revision creation a bit less patchy
169 2010-05-24T18:55:55 <TheSheep> ThomasWaldmann: I wonder, do we keep things like trying to convert _ to spaces if the page doesn't exist?
170 2010-05-24T18:56:09 <TheSheep> ThomasWaldmann: in moin 2.0
171 2010-05-24T19:04:06 <ThomasWaldmann> well, would be good if it could be done (for links from outside)
172 2010-05-24T19:04:43 <ThomasWaldmann> for now we can have a TODO for that :)
173 2010-05-24T19:05:40 *** DmitryAndreev
174 2010-05-24T19:07:06 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8238:0b0994f20aee 2.0-dev/MoinMoin/storage/backends/fs2.py: Item._backend and Revision._backend are initialized by base classes
175 2010-05-24T19:29:34 * dreimark debugs draft
176 2010-05-24T19:38:14 * ThomasWaldmann tries a history from index
177 2010-05-24T19:39:42 *** CIA-47
178 2010-05-24T19:54:44 *** CIA-47
179 2010-05-24T19:58:21 *** kursor
180 2010-05-24T20:14:36 <ThomasWaldmann> great, RouterBackend is misdesigned :|
181 2010-05-24T21:14:39 <ThomasWaldmann> hmm, history is a bit troublesome
182 2010-05-24T21:15:07 <ThomasWaldmann> if i have a index per backend, i also have history per backend
183 2010-05-24T21:16:55 <ThomasWaldmann> so, for global history, I'ld need to mix it together
184 2010-05-24T21:19:34 <ThomasWaldmann> also, names are only the part of the name the backend is managing (without mountpoint)
185 2010-05-24T21:29:03 *** ei-grad
186 2010-05-24T21:46:29 <dreimark> cache arena with pickle is to expensive if there are many pages as draft saved
187 2010-05-24T21:50:57 <dreimark> once loaded it stays in memory
188 2010-05-24T21:51:10 <valeuf> did you see test_html.py in EmeraldTRee ?
189 2010-05-24T21:51:39 <valeuf> same style of test could be use to test the output of the html_out converter ?
190 2010-05-24T21:52:08 *** ei-grad
191 2010-05-24T21:53:53 <ThomasWaldmann> dreimark: what does "it stays in memory" mean exactly?
192 2010-05-24T21:54:04 <valeuf> also, there is html.py in EmeraldTree, to convert html to an EmeraldTree
193 2010-05-24T21:54:14 <valeuf> maybe I can use it for my html_in converter ?
194 2010-05-24T21:56:37 <dreimark> http://paste.pocoo.org/show/217797/
195 2010-05-24T21:56:48 <dreimark> pickle.loads(data)
196 2010-05-24T21:57:09 <dreimark> and another one increases the amount of memory used
197 2010-05-24T21:57:31 <ThomasWaldmann> ?
198 2010-05-24T21:58:15 <dreimark> if i have that on a python shell
199 2010-05-24T21:58:34 <dreimark> always whith every call of pickle.loads(data)
200 2010-05-24T21:58:48 <dreimark> i get more and more memory used
201 2010-05-24T21:59:17 <ThomasWaldmann> that's not the point
202 2010-05-24T21:59:50 <dreimark> when will the memory cleared it uses
203 2010-05-24T22:00:12 <ThomasWaldmann> when the object gets garbage collected
204 2010-05-24T22:01:07 <ThomasWaldmann> (13:46) < ThomasWal> please check where it loads the draft cache and whether it can be garbage collected
205 2010-05-24T22:02:16 <dreimark> i know readed that
206 2010-05-24T22:05:41 <dreimark> i try to understand how i can garbage collect it
207 2010-05-24T22:10:15 <dreimark> http://hg.moinmo.in/moin/1.9/file/203afc609f8a/MoinMoin/caching.py#l280
208 2010-05-24T22:18:44 <ThomasWaldmann> you need to look at the called of content(), what it does with the returned value
209 2010-05-24T22:19:15 <ThomasWaldmann> and an object is garbage collected when gc runs and it is not referenced any more
210 2010-05-24T22:27:10 <ThomasWaldmann> *caller
211 2010-05-24T22:27:36 * dreimark reads http://www.electricmonk.nl/log/2008/07/07/python-destructor-and-garbage-collection-notes/
212 2010-05-24T22:28:23 <ThomasWaldmann> do we have __del__?
213 2010-05-24T22:31:14 <dreimark> we don't but some of our support libs
214 2010-05-24T23:05:44 <valeuf> I made some experiences with Xpath, but it is not really working well
215 2010-05-24T23:05:45 <valeuf> http://www.pastie.org/975182
216 2010-05-24T23:14:22 <ThomasWaldmann> see your page
217 2010-05-24T23:17:19 <valeuf> ThomasWaldmann: sorry, the default theme of pastie is not that goo
218 2010-05-24T23:17:21 <valeuf> *good
219 2010-05-24T23:17:26 <valeuf> I put the source on the wiki
220 2010-05-24T23:17:50 <dreimark> valeuf: usually we use paste.pocoo.org
221 2010-05-24T23:17:58 <dreimark> what is wrong with it?
222 2010-05-24T23:17:59 <valeuf> for the first problem, it was what I was talking about with xorAxAx
223 2010-05-24T23:18:48 <valeuf> dreimark: no problem, I will change my vim script to use paste.pocoo.org
224 2010-05-24T23:19:19 <xorAxAx> valeuf: doesnt load
225 2010-05-24T23:19:34 <xorAxAx> ah, now it does
226 2010-05-24T23:19:38 <valeuf> xorAxAx: load what ?
227 2010-05-24T23:20:01 <valeuf> you can check directly : http://moinmo.in/ValentinJaniaut/GSoC/ProblemsAndQuestion instead of the pastie
228 2010-05-24T23:20:31 <xorAxAx> well, that test is nearly useless
229 2010-05-24T23:20:36 <xorAxAx> its not strict enough
230 2010-05-24T23:21:05 <valeuf> xorAxAx: yep I was writting some comment about it
231 2010-05-24T23:21:17 <valeuf> but I am not sure by approach is right
232 2010-05-24T23:22:21 <valeuf> xorAxAx: to make it more strict, first of all I should care about the order of the tags,
233 2010-05-24T23:22:52 <valeuf> xorAxAx: but I think that iterating with Xpath on the tree is just stupid
234 2010-05-24T23:23:02 <xorAxAx> yes, your code is a bit of that :)
235 2010-05-24T23:23:08 <xorAxAx> do you know how xpath works?
236 2010-05-24T23:23:42 <valeuf> however div/p/sub does not
237 2010-05-24T23:23:43 <valeuf> work
238 2010-05-24T23:25:47 <valeuf> I just get an empty list
239 2010-05-24T23:29:39 <valeuf> I will add some unit test in emerald tomorrow, to see if XPath stuff are working correctly
240 2010-05-24T23:56:23 <ThomasWaldmann> valeuf: i added some unit tests there recently