1 2010-05-09T00:59:48 *** ronny
2 2010-05-09T00:59:49 *** elmo
3 2010-05-09T00:59:49 *** dennda
4 2010-05-09T00:59:49 *** aa_
5 2010-05-09T01:00:36 *** ronny
6 2010-05-09T01:00:37 *** aa_
7 2010-05-09T01:00:37 *** dennda
8 2010-05-09T01:00:37 *** elmo
9 2010-05-09T01:00:44 *** ronny
10 2010-05-09T01:00:44 *** elmo
11 2010-05-09T01:00:44 *** dennda
12 2010-05-09T01:00:44 *** aa_
13 2010-05-09T01:01:13 *** ronny
14 2010-05-09T01:01:13 *** aa_
15 2010-05-09T01:01:13 *** dennda
16 2010-05-09T01:01:13 *** elmo
17 2010-05-09T03:37:43 <ronny> sup
18 2010-05-09T03:37:51 <ronny> oh, wtf
19 2010-05-09T03:37:53 <ThomasWaldmann> http://hg.moinmo.in/moin/2.0-dev/file/d18e362af49e/MoinMoin/wsgiapp.py#l138
20 2010-05-09T03:38:04 <ronny> (i just saw the time)
21 2010-05-09T03:38:05 <ThomasWaldmann> this is the current way how middleware is set up
22 2010-05-09T03:38:36 <ThomasWaldmann> it is the acl checking middleware, but the problem is similar
23 2010-05-09T03:40:42 <ThomasWaldmann> (and when you look at MoinMoin.storage.backends.acl, you see it is a bit strange, doing sometimes proxying, sometimes inheritance, ...)
24 2010-05-09T03:43:27 <ronny> too complex for me atm, im too tired to propperly get an mindmodel of whats going on
25 2010-05-09T03:43:54 <ThomasWaldmann> yeah, I guess me too :)
26 2010-05-09T03:44:47 <ThomasWaldmann> would appreciate some more concrete tipps, because this is currently blocking me with the indexing middleware
27 2010-05-09T03:45:07 <ThomasWaldmann> the sql stuff more or less works already, also the kvstore
28 2010-05-09T03:46:25 <ronny> ThomasWaldmann: what ican say is that decoration/proxying feels just wrong for the thing
29 2010-05-09T03:46:32 <ronny> its a composition issue
30 2010-05-09T03:46:37 <ThomasWaldmann> btw, i did the serialization/unserialization stuff as mixin for the backend/item/revision base class
31 2010-05-09T03:47:08 <ronny> poke me after some sleep and i'll take a deeper look
32 2010-05-09T03:47:20 <ThomasWaldmann> and that is the point where the indexing middleware as i did it first failed: the item unserializer didn't call the method of the imw, but just the stuff below it
33 2010-05-09T03:47:58 <ronny> middleware is just plain wrong for those issues
34 2010-05-09T03:48:12 <ronny> i'll go into more detail after sleep
35 2010-05-09T03:48:17 <ronny> night
36 2010-05-09T03:49:22 <ThomasWaldmann> gn
37 2010-05-09T03:49:40 * ThomasWaldmann leaves hackspace and drives home :)
38 2010-05-09T07:01:53 *** kursor
39 2010-05-09T09:16:26 <dreimark> moin
40 2010-05-09T09:57:17 <dreimark> waldi: do we have some tidy functionality in the dom tree stuff?
41 2010-05-09T09:57:52 <dreimark> which can fix e.g. <td width=90% colspan=2"><font size="-1">
42 2010-05-09T09:58:05 <dreimark> s/colspan=2"/colspan="2"
43 2010-05-09T09:58:16 <dreimark> and quote the others?
44 2010-05-09T10:19:10 <waldi> dreimark: this is no valid markup
45 2010-05-09T10:32:41 <dreimark> i know
46 2010-05-09T10:34:17 <dreimark> this markup is incredible broken, I wonder that tidylib didn't crash
47 2010-05-09T11:01:06 <dreimark> how cab I overwrite xmlrpc_putAttachment in wikiconfig?
48 2010-05-09T11:01:15 <dreimark> s/cab/can/
49 2010-05-09T11:02:05 *** kursor
50 2010-05-09T11:13:02 <dreimark> TheSheep: non theming dependent plugins need to work with the new plugin system too
51 2010-05-09T11:13:24 <dreimark> e.g. xmlrpc, auth ...
52 2010-05-09T11:44:04 <ThomasWaldmann> moin
53 2010-05-09T11:50:55 <dreimark> hi TheSheep
54 2010-05-09T11:50:58 <dreimark> ThomasWaldmann:
55 2010-05-09T11:51:34 <dreimark> do we have plans for edit and event log ?
56 2010-05-09T11:51:55 <ThomasWaldmann> a tree doesn't need cleaning and there is no html > dom converter yet
57 2010-05-09T11:52:26 <dreimark> by the mass operations i currently do I just wish to separate where new entries are logged
58 2010-05-09T11:53:14 <ThomasWaldmann> and my current thoughts about event log is that it is rather superfluous
59 2010-05-09T11:54:01 <ThomasWaldmann> and edit-log will likely be replaced by revision metadata and an index on that
60 2010-05-09T11:54:38 <dreimark> xmlrpc or cli operations should be abled to have a separate edit-log
61 2010-05-09T11:56:10 <dreimark> rc becomes unreadable on a mass creation of items
62 2010-05-09T12:04:15 <ThomasWaldmann> rc is expected to reflect changes, so if you do mass changes, you'll have a massive rc
63 2010-05-09T12:08:16 <ronny> moin
64 2010-05-09T12:09:19 <dreimark> i know but it also could show a summary and could have plugins for detailed view
65 2010-05-09T12:10:01 <dreimark> so it could tell 1000 uploads at page foo and have a plugin to show the granularity
66 2010-05-09T12:11:38 <ThomasWaldmann> I'll think about that after the base work has been done
67 2010-05-09T12:13:04 <dreimark> :)
68 2010-05-09T12:23:18 <ThomasWaldmann> TheSheep: ronny: if you want to look again:
69 2010-05-09T12:24:16 <ThomasWaldmann> the problem is how to setup storage backends and do acl checks (currently middleware, have a look at src) and catch the metadata/data for indexing
70 2010-05-09T12:24:34 <ThomasWaldmann> and, if possible, without doing widespread changes to all backends
71 2010-05-09T12:27:05 <ThomasWaldmann> the acl middleware currently looks a bit strange btw. for backend it does proxying, for other classes it uses inheritance ...
72 2010-05-09T12:29:56 <TheSheep> ThomasWaldmann: I was thinking about it a little, and I think that inheritance is the way to go
73 2010-05-09T12:30:54 <TheSheep> ThomasWaldmann: you can have something like def mix(*classes): class Class(*classes): pass; return Class
74 2010-05-09T12:30:59 <TheSheep> (in pseudocode)
75 2010-05-09T12:31:29 <TheSheep> then you can mix classes at runtime
76 2010-05-09T12:32:14 <TheSheep> you can't really use * with classes :(
77 2010-05-09T12:35:25 <valeuf> do you think that adding general converter in specific converter like html_out is a bad things ?
78 2010-05-09T12:35:49 <valeuf> I feel like it could be cood ta have one converter with all the comments about is working generally a converter
79 2010-05-09T12:36:00 <valeuf> so people would be able to use it as an example later
80 2010-05-09T12:50:53 <ThomasWaldmann> hi valeuf
81 2010-05-09T12:51:02 <valeuf> ThomasWaldmann: hi :)
82 2010-05-09T12:51:27 <ThomasWaldmann> can you rephrase? i don't think i got you.
83 2010-05-09T12:51:39 <valeuf> ThomasWaldmann: you are right, it is not really clear
84 2010-05-09T12:52:20 <valeuf> ThomasWaldmann: thanks to waldi, I am reading the documentation of the Tree Output Formatter http://moinmo.in/BastianBlank/TreeOutputFormatter/Documentation
85 2010-05-09T12:53:22 <valeuf> Actually, when I was reading the source code of a basic converter like html_out.py , I was not able to figure easily many basic things about converter
86 2010-05-09T12:54:09 <ronny> ThomasWaldmann: i really think the composition is kind of wrong there
87 2010-05-09T12:54:11 <ThomasWaldmann> well, likely the src docs need improvements
88 2010-05-09T12:54:42 <valeuf> for example, it was not so clear that __call__ will be the function called by the converter factory to run the conversion
89 2010-05-09T12:55:30 <ThomasWaldmann> valeuf: how about creating a docstring-adding changeset with what you've found out and let it proofread by waldi
90 2010-05-09T12:55:40 <valeuf> ThomasWaldmann: great :)
91 2010-05-09T12:55:49 <valeuf> Anyway I started to do it :p
92 2010-05-09T12:55:57 <valeuf> It is helpful for me
93 2010-05-09T12:58:35 <ronny> ThomasWaldmann: its actually rather tricky to guess how to get those things in, i think composition at the level of the base-classes might be the best way
94 2010-05-09T12:58:56 <ronny> (but thats just massivel different from the curent model)
95 2010-05-09T12:59:20 <ThomasWaldmann> TheSheep: http://www.gamedev.net/community/forums/topic.asp?topic_id=445037
96 2010-05-09T12:59:33 <ThomasWaldmann> but, "new" seems deprecated
97 2010-05-09T13:00:47 <ThomasWaldmann> ronny: how would that look like for acl checks?
98 2010-05-09T13:01:20 <ronny> ThomasWaldmann: well, the base class for the backends would call out to the acl thing instead of a wrapper/decorator
99 2010-05-09T13:02:31 <ronny> ThomasWaldmann: alternatively one could make backends more fit to be wrapped/decorated, but i cant make up my mind on that till i know the basic asumptions for interacting with backends
100 2010-05-09T13:02:42 <ronny> (i really dont have those in my mind atm)
101 2010-05-09T13:03:07 <ThomasWaldmann> well, one point of the acl middleware was doing it very safely
102 2010-05-09T13:03:48 <ThomasWaldmann> that means if the acl middleware sits in between wiki code and backend, you can't do wrong with acl checks
103 2010-05-09T13:03:49 <TheSheep> ThomasWaldmann: when you need to inset something between internal calls of the objects, then you know you don't have enough layers
104 2010-05-09T13:04:00 <ThomasWaldmann> assuming that the middleware itself is correct
105 2010-05-09T13:15:01 <ThomasWaldmann> TheSheep: > Test = type(
106 2010-05-09T13:15:02 <ThomasWaldmann> > 'Test', (FirstBase, SecondBase), attr)
107 2010-05-09T13:15:10 <TheSheep> eww
108 2010-05-09T13:15:49 <TheSheep> ThomasWaldmann: that's not the solution to the architectural problem
109 2010-05-09T13:16:03 <TheSheep> ThomasWaldmann: you don't want to put things between internal calls
110 2010-05-09T13:16:51 <ThomasWaldmann> that was about how to dynamically create classes
111 2010-05-09T13:34:45 <ThomasWaldmann> (== your "mix" function from above)
112 2010-05-09T13:43:06 <TheSheep> yeah
113 2010-05-09T13:43:19 <TheSheep> but it seems dirty
114 2010-05-09T13:44:11 <TheSheep> it seems that the storage object should be divided into a higher and lower level parts
115 2010-05-09T14:10:29 <CIA-55> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5667:f7f9d35f9d23 1.9/MoinMoin/userprefs/oid.py: fix associating OpenID identity with user, wasn't adapted to werkzeug yet (thanks to Paul Boddie for the patch)
116 2010-05-09T17:18:43 *** ei-grad
117 2010-05-09T18:13:06 *** ei-grad
118 2010-05-09T22:53:43 <CIA-55> Reimar Bauer <rb.proj AT googlemail DOT com> default * 5668:e77e7448ae1f 1.9/MoinMoin/converter/text_html_text_moin_wiki.py: text_html_text_moin_wiki: fixed index error for a width="". (MoinMoinBugs/GUI_convertor_list_index_out_of_range)
