1 2010-05-23T01:11:47 * ThomasWaldmann takes another attempt at metadata indexing infrastructure, this time with Mixins
2 2010-05-23T01:28:34 <dreimark> gn
3 2010-05-23T01:28:51 * dreimark continues tom with memory debugging
4 2010-05-23T08:54:45 *** diofeher
5 2010-05-23T10:19:07 <dreimark> moin
6 2010-05-23T10:19:17 * dreimark needs some tips for a farm setup
7 2010-05-23T10:19:59 <dreimark> the most interesting question is if i need to setup for each farm wiki a WSGIScriptAlias?
8 2010-05-23T10:20:32 <dreimark> WSGIScriptAlias / /some/path/moin.wsgi
9 2010-05-23T10:20:59 <dreimark> or if it is enough to setup / and the farmconfig manages then the wikis
10 2010-05-23T10:24:08 * dreimark should have readed the docs
11 2010-05-23T10:25:33 <dreimark> looks like i can use a regex :)
12 2010-05-23T10:25:55 <dreimark> tht makes it easy
13 2010-05-23T11:35:53 <ThomasWaldmann> moin
14 2010-05-23T12:03:49 <TheSheep> ThomasWaldmann: you want to try and make moin use werkzeug's routing?
15 2010-05-23T12:14:28 <ThomasWaldmann> yes.
16 2010-05-23T12:16:35 <TheSheep> in moin-2.0-dev?
17 2010-05-23T12:17:09 <ThomasWaldmann> how long it'll take?
18 2010-05-23T12:19:59 <TheSheep> I don't know, but I guess it doesn't require both of us present
19 2010-05-23T12:22:27 <ThomasWaldmann> I guess I have some time for hacking today (I was still working on metadata indexing integration, backend datastructures and init.)
20 2010-05-23T12:23:01 <TheSheep> ThomasWaldmann: maybe you could just clone me a branch and I would try hacking in it, and you'd work on your hacks
21 2010-05-23T12:23:20 <ThomasWaldmann> ok, -rest?
22 2010-05-23T12:23:49 <TheSheep> sounds good
23 2010-05-23T12:26:00 <ThomasWaldmann> http://hg.moinmo.in/moin/2.0-dev-rest/
24 2010-05-23T12:26:11 <TheSheep> thanks
25 2010-05-23T12:26:31 <TheSheep> I'm happy because i rpped out over 300 lines from Hatta today :)
26 2010-05-23T12:26:35 <TheSheep> ripped
27 2010-05-23T12:27:18 <ThomasWaldmann> ripped out = copied for moin or ripped out from hatta with optimization?
28 2010-05-23T12:27:46 <TheSheep> ripped out from hatta, so its sources are 300 lines shorter
29 2010-05-23T12:28:07 <ThomasWaldmann> ok, that's quite much, considering size of hatta
30 2010-05-23T12:28:09 <TheSheep> rewrote the rss feed to use werkzeug.contrib.atom
31 2010-05-23T12:28:22 <TheSheep> 10%
32 2010-05-23T12:30:07 <ThomasWaldmann> for the backend stuff, i have modified fs2 yesterday to have a hierarchy of classes (with mixins)
33 2010-05-23T12:30:36 <ThomasWaldmann> and the final classes are named like the previous ones
34 2010-05-23T12:30:57 <ThomasWaldmann> seems to work so far
35 2010-05-23T12:32:14 <ThomasWaldmann> likely will need it to have request somehow (create backend per request), that's not the case yet
36 2010-05-23T12:35:29 <TheSheep> backend per request would be bad :(
37 2010-05-23T12:35:57 <TheSheep> especially if it's a database backend, each with its own db connection
38 2010-05-23T12:35:58 <ThomasWaldmann> acl stuff needs request.user.name
39 2010-05-23T12:36:22 <TheSheep> maybe it can get the request object as a parameter
40 2010-05-23T12:36:26 <ThomasWaldmann> indexing stuff needs index db location / connection
41 2010-05-23T12:36:26 <ronny_> ThomasWaldmann: use threadlocal context?
42 2010-05-23T12:37:23 <ThomasWaldmann> iirc wsgi doesn't guarantee 1 thread per request
43 2010-05-23T12:38:02 <TheSheep> yup
44 2010-05-23T12:43:52 <ronny_> ThomasWaldmann: werkzeug has tools to pass that kind of object around
45 2010-05-23T15:29:19 <ThomasWaldmann> http://werkzeug.pocoo.org/documentation/0.6.2/local.html
46 2010-05-23T15:40:49 <ThomasWaldmann> TheSheep: i am updating werkzeug to 0.6.2 in moin/2.0-dev, OK?
47 2010-05-23T16:06:17 <dreimark> ThomasWaldmann: we use global in i18n in 1.9
48 2010-05-23T16:07:28 <dreimark> and on some other places
49 2010-05-23T16:08:55 <ThomasWaldmann> i know
50 2010-05-23T16:10:58 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8219:aea8221ad7d3 2.0-dev/MoinMoin/ (3 files in 2 dirs): (log message trimmed)
51 2010-05-23T16:10:58 <CIA-47> reverse indexing middleware and some related changesets (list below) - Proxy approach doesn't work
52 2010-05-23T16:10:58 <CIA-47> changeset: 8214:fc1235696cf9
53 2010-05-23T16:10:58 <CIA-47> user: Thomas Waldmann <tw AT waldmann-edv DOT de>
54 2010-05-23T16:10:58 <CIA-47> date: Sun May 16 21:06:55 2010 +0200
55 2010-05-23T16:10:59 <CIA-47> summary: add an object proxy class (MoinMoin.support.objectproxy)
56 2010-05-23T16:11:00 <CIA-47> changeset: 8215:679353aade79
57 2010-05-23T16:11:04 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8220:dff2ea3c78d3 2.0-dev/MoinMoin/support/werkzeug/ (40 files in 4 dirs): update bundled werkzeug to 0.6.2 release
58 2010-05-23T16:12:37 <TheSheep> ThomasWaldmann: thanks
59 2010-05-23T16:12:52 <TheSheep> ThomasWaldmann: I'm still analyzing the code %)
60 2010-05-23T16:19:15 <ThomasWaldmann> moin is twisted
61 2010-05-23T16:20:35 <ThomasWaldmann> TheSheep: maybe we can start with putting request into a context local ADDITIONALLY
62 2010-05-23T17:45:44 *** DmitryAndreev
63 2010-05-23T18:59:07 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8221:9e245b5a9392 2.0-dev/wikiconfig.py: working default backend uri in wikiconfig (using fs2, fs backend is broken currently)
64 2010-05-23T18:59:39 <ThomasWaldmann> if someone is bored, he could look at what's wrong with fs backend
65 2010-05-23T19:18:22 <ThomasWaldmann> i currently have some issues with type inconsistencies within the metadata
66 2010-05-23T19:18:36 <ThomasWaldmann> e.g. str vs unicode
67 2010-05-23T19:19:30 <ThomasWaldmann> I am thinking about dropping str (because for most DBs, unicode gets encoded to utf-8 anyway and then for ascii strs, output is the same anyway)
68 2010-05-23T19:23:10 <ThomasWaldmann> and I don't think we'll use str to store non-text data or have bytestrings with unknown encoding
69 2010-05-23T19:25:52 <ThomasWaldmann> rev meta 'edit_log_addr': '::1'
70 2010-05-23T19:26:08 <ThomasWaldmann> werkzeug 0.6.2 with ipv6...
71 2010-05-23T21:04:21 *** DmitryAndreev
72 2010-05-23T21:14:21 *** diofeher
73 2010-05-23T21:53:58 *** diofeher
74 2010-05-23T21:55:12 * ThomasWaldmann fixes types in fs19
75 2010-05-23T22:06:11 *** diofeher
76 2010-05-23T22:28:06 <xorAxAx> valeuf: so whats the problem with xpath and tests?
77 2010-05-23T22:59:47 <ThomasWaldmann> xorAxAx: could you review backends whether sth is holding us back from uuids for items?
78 2010-05-23T23:04:15 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8222:2c1292f1b180 2.0-dev/MoinMoin/storage/backends/fs19.py: fs19: add comment about bad acl migration, fix code typo, fix pep8
79 2010-05-23T23:05:06 <xorAxAx> ThomasWaldmann: which problems do you expect?
80 2010-05-23T23:07:20 *** DmitryAndreev
81 2010-05-23T23:08:58 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8223:845003fe60c1 2.0-dev/MoinMoin/storage/backends/fs19.py: fs19: use unicode for text-like metadata values
82 2010-05-23T23:11:47 <ThomasWaldmann> xorAxAx: not really a (code) problem, rather that some stuff might be expected to rather deal with names
83 2010-05-23T23:12:08 <ThomasWaldmann> or does its own uuid stuff internally in some special way
84 2010-05-23T23:20:26 <ThomasWaldmann> the current item name (==latest revision's name) could be in item_meta['name'] and the revision's name is in rev_meta['name']
85 2010-05-23T23:24:58 <diofeher> good night!
86 2010-05-23T23:25:20 <diofeher> hey, i was wondering if it's best to rewrite a new theme now or to fix the old default
87 2010-05-23T23:30:55 <CIA-47> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 8224:e2577408c0eb 2.0-dev/MoinMoin/wikiutil.py: (log message trimmed)
88 2010-05-23T23:30:55 <CIA-47> fs19: fix metadata value list/tuple handling
89 2010-05-23T23:30:55 <CIA-47> It doesn't depend on the length of the values whether we want a tuple or a
90 2010-05-23T23:30:55 <CIA-47> single value, but on the key.
91 2010-05-23T23:30:55 <CIA-47> For # key (comments and pseudo-processing stuff), we want a tuple, so we can
92 2010-05-23T23:30:56 <CIA-47> reproduce it 1:1.
93 2010-05-23T23:30:57 <CIA-47> For ACLs made of multiple #acl lines, we want to join them with a blank.
94 2010-05-23T23:31:24 <ThomasWaldmann> diofeher: i guess starting from scratch is better
95 2010-05-23T23:31:57 <ThomasWaldmann> of course you can peek into the old stuff and selectively copy some stuff, if you find it useful
96 2010-05-23T23:35:26 <diofeher> ThomasWaldmann: nice... so i will replace modernized... blue is preferred to use in theme?
97 2010-05-23T23:36:52 <diofeher> i will design some themes and then show up in channel, then you all decide the way i have to follow :)
98 2010-05-23T23:44:32 <diofeher> and i think georgia is a good font to use
99 2010-05-23T23:49:18 <ThomasWaldmann> diofeher: i would suggest you stay more connected here
100 2010-05-23T23:50:10 <ThomasWaldmann> and communicate MUCH more often than you did until now
101 2010-05-23T23:51:09 <ThomasWaldmann> and before you even thing about colours and fonts, create the basic infrastructure for the theme
102 2010-05-23T23:51:13 <ThomasWaldmann> *think
103 2010-05-23T23:53:45 <ThomasWaldmann> and later well see what we can do about the look and feel
104 2010-05-23T23:53:59 *** diofeher
105 2010-05-23T23:57:36 *** diofeher
106
