1 2010-08-29T00:35:59 *** RogerHaase
2 2010-08-29T01:15:15 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9780:e1059798c500 2.0-dev/MoinMoin/ (26 files in 18 dirs): move (unprotected_)storage to flaskg
3 2010-08-29T10:55:27 *** grzywacz
4 2010-08-29T11:40:05 <ThomasWaldmann> moin
5 2010-08-29T12:28:16 * ThomasWaldmann updated http://code.google.com/p/google-summer-of-code-2010-moinmoin/
6 2010-08-29T15:58:58 *** RogerHaase
7 2010-08-29T16:27:19 *** jdwcornell
8 2010-08-29T16:30:30 <ThomasWaldmann> re
9 2010-08-29T16:35:12 <jdwcornell> question about moin2 hg backend vision: is it a goal/vision/intention to allow hg clone/push/pull from a moin2 instance with item names instead of <whatever the hex identifier is called>
10 2010-08-29T16:36:29 <jdwcornell> so, Home, rst, moin, creole and docbook as files in the (remotely cloned) repo instead of 8cf04a9734132302f96da8e113e80ce5{,.rev}, d674dfcd8b4db6762bcb3667316d3bb9{,.rev}, etc
11 2010-08-29T16:37:34 <ThomasWaldmann> there wasn't recent work on the hg backend
12 2010-08-29T16:38:18 <jdwcornell> I guess another way to ask the question would be "what's the architectural vision of the moin2 folks" on that matter?
13 2010-08-29T16:38:58 <ThomasWaldmann> in general, the backends have to behave in some common way on the moin api side
14 2010-08-29T16:39:09 <jdwcornell> I'm interested in that kind of functionality, and could be convinced to implement it, but I've run into projects before where the core developers have declined patches because it didn't meet with the vision
15 2010-08-29T16:39:48 <ThomasWaldmann> the concept of the moin storage api is to have some storage which has items
16 2010-08-29T16:39:56 <ThomasWaldmann> items have metadata and revisions
17 2010-08-29T16:40:04 <ThomasWaldmann> revisions have metadata and data
18 2010-08-29T16:40:08 <jdwcornell> yup - I'm talking about some kind of "frontend" exposure. Kind of like bitbucket allows you to edit your wiki on there with a cloned hg repo
19 2010-08-29T16:41:01 <ThomasWaldmann> the problem with manually editing is that data and metadata is not that easily put into editable files
20 2010-08-29T16:41:32 <jdwcornell> yeah the more I look into it (I'm just at the tip of my investigation here) I'm seeing that as a barrier to what I'm thinking
21 2010-08-29T16:41:58 <ThomasWaldmann> and the problem with using item names as file names is that this does not work in general
22 2010-08-29T16:42:48 <jdwcornell> due to spaces, punctuation and other "odd" characters, right?
23 2010-08-29T16:43:15 <ThomasWaldmann> yes, and due to different operating system specialities
24 2010-08-29T16:44:01 <ThomasWaldmann> also, a more recent (and not fully finished) development will maybe make backend development easier
25 2010-08-29T16:44:20 <jdwcornell> gotcha - makes sense.
26 2010-08-29T16:44:40 <ThomasWaldmann> it might be an option that an upper layer of storage infrastructure (maybe the router layer) translates names to uuids
27 2010-08-29T16:45:00 <jdwcornell> I just jumped in this morning (long time moinmoin user though) by cloning http://hg.moinmo.in/moin/2.0-dev
28 2010-08-29T16:45:01 <ThomasWaldmann> uuids of items would never change, even when their name is changing
29 2010-08-29T16:46:28 <ThomasWaldmann> if there were some easy way to put stuff into files, edit them and push them back, I'ld be open to discuss it
30 2010-08-29T16:46:38 <ThomasWaldmann> but currently I see rather trouble there
31 2010-08-29T16:47:41 <jdwcornell> ok - I'll stew on this some more. agreed re: trouble. where I'm coming from is a perspective / lens of sometimes wanting to click around the GUI (browser) and sometimes liking my emacs interface to a filesystem
32 2010-08-29T16:47:57 <ThomasWaldmann> for people disliking the html edit form, there could be some sort of clientside script maybe
33 2010-08-29T16:48:55 <ThomasWaldmann> the script would just fetch the (single revision) data from the wiki, feed editor, write it back when wanted
34 2010-08-29T16:49:55 <jdwcornell> now that gives me visions of a moin fuse plugin :)
35 2010-08-29T16:50:08 <ThomasWaldmann> that's much easier than trying to have some whole wiki in some super-easy editable storage that works for everyone
36 2010-08-29T16:51:25 <jdwcornell> yup, and it could even be partially crippled in terms of which items it lets you edit. say, ones with "undesirable" names wouldn't be accessible
37 2010-08-29T16:51:41 <ThomasWaldmann> btw, I'ld like to get some more feedback on the uuid ideas (there wasn't too much of it yet).
38 2010-08-29T16:51:43 <jdwcornell> depending on the translation layer
39 2010-08-29T16:53:08 <jdwcornell> where would I find more articulation of the UUID ideas, and what's the preferred method of feedback with this crowd?
40 2010-08-29T16:53:30 <ThomasWaldmann> i guess the preferred method is here and on the wiki
41 2010-08-29T16:53:45 <jdwcornell> k
42 2010-08-29T16:54:57 <ThomasWaldmann> i guess a full text search for uuid on the wiki gives some results
43 2010-08-29T16:55:37 <jdwcornell> yup doing that now. I'll need to dig deeper into the code too before I can provide useful commentary
44 2010-08-29T16:55:52 <ThomasWaldmann> if you look into the fs and fs2 backends, you see they do something similar internally
45 2010-08-29T16:56:05 <ThomasWaldmann> with cdb / sqla
46 2010-08-29T16:56:34 <ThomasWaldmann> if you can't store the real name onto the fs, you want something like that
47 2010-08-29T16:56:55 <ThomasWaldmann> but there is no reason why every backend has to implement its own way to do that
48 2010-08-29T16:59:14 <jdwcornell> thanks for the feedback and background. much appreciated.
49 2010-08-29T17:00:16 <ThomasWaldmann> i am currently working in the wsgi/moin request cleanup area
50 2010-08-29T17:00:36 <ThomasWaldmann> but after that, I guess I'll do more storage related work again
51 2010-08-29T17:01:18 <ThomasWaldmann> i didn't do too much with the mercurial and sqlalchemy based backends recently,
52 2010-08-29T17:01:54 <ThomasWaldmann> so there was always the question if these ideas are compatible with those (esp. hg) and who implements the changes there
53 2010-08-29T17:03:42 <ThomasWaldmann> jdwcornell: btw, did you play with the test wiki?
54 2010-08-29T17:07:35 *** kursor
55 2010-08-29T17:08:52 <jdwcornell> I think an important question to know the answer to is "what features must be implemented by the backend"? For instance, hg is *really* good at natively storing revisions. But the FS ones, sql ones, etc don't do that natively and must be "simulated" at the point of the backend's exposure to the rest of the stack. But Hg isn't particularly well suited for, say full text search, like a database like Postgres or Oracle which have it bu
56 2010-08-29T17:08:52 <jdwcornell> in. So do you expect the storage backend to expose versioning but not full text search? where exactly in the stack is it appropriate to expose versioning, full text search, etc? always at the storage backend point? somewhere else? has this been "decided/discussed" already, per se?
57 2010-08-29T17:09:33 <jdwcornell> I've been messing around with python wikiserver.py in the tip of http://hg.moinmo.in/moin/2.0-dev
58 2010-08-29T17:10:10 <jdwcornell> a related question might be "why bother with all those different backends"? sysadmin flexibility?
59 2010-08-29T17:11:21 <jdwcornell> I'm not questioning the decision to do so; indeed I can see lots of value in that approach. understanding the rationales, though, helps understand the existing architecture, and where best to integrate new features
60 2010-08-29T17:11:38 <ThomasWaldmann> search isn't really implemented yet
61 2010-08-29T17:12:11 <ThomasWaldmann> there are some remainders of moin 1.9 (slow builtin search, xapian indexed search), but they won't work without bigger changes
62 2010-08-29T17:12:34 <ThomasWaldmann> what's new is that there will be a metadata index at the router layer above the storage,
63 2010-08-29T17:13:01 <ThomasWaldmann> so if you just search for all items with mimetype image/png, that layer will be able to do that
64 2010-08-29T17:13:24 <jdwcornell> gotcha - so that makes it so the search can be implemented in the storage layer if it is more efficient to do so (like Pg, ora full text)
65 2010-08-29T17:13:48 <ThomasWaldmann> and because of different capabilities, it is unlikely that we'll use specific search stuff in the backend itself
66 2010-08-29T17:15:09 <jdwcornell> but versioning is expected to be implemented in the backend?
67 2010-08-29T17:15:18 <ThomasWaldmann> versioning has to be implemented by backend, that is part of the api
68 2010-08-29T17:15:57 <ThomasWaldmann> also storing metadata AND data
69 2010-08-29T17:16:05 <jdwcornell> heh: http://moinmo.in/ThomasWaldmann/Moin2BrainDump?highlight=%28searching%29
70 2010-08-29T17:16:52 <ThomasWaldmann> i currently think that there will be 2 storage apis
71 2010-08-29T17:17:06 <ThomasWaldmann> one for the "it really stores data" backends
72 2010-08-29T17:17:29 <ThomasWaldmann> one for the router, which knows full item names, which does indexing etc.
73 2010-08-29T17:17:55 <ThomasWaldmann> the backend can be rather dump
74 2010-08-29T17:18:27 <ThomasWaldmann> just "store this stuff" and "list items stored" functionality
75 2010-08-29T17:19:04 <ThomasWaldmann> the upper layer (router) will use that to build an index and offer searching, name/uuid mapping etc.
76 2010-08-29T17:19:22 <ThomasWaldmann> dumb*
77 2010-08-29T17:20:21 <ThomasWaldmann> the point for offering different backends is different tastes and different needs
78 2010-08-29T17:20:48 <ThomasWaldmann> some people don't like to maintain sql database servers, they just want something simple using the filesystem
79 2010-08-29T17:21:06 <ThomasWaldmann> and others just want it the other way
80 2010-08-29T17:21:15 <jdwcornell> the question would then be: what does an SQL server get you, in the context of MoinMoin?
81 2010-08-29T17:21:28 <jdwcornell> backup managability?
82 2010-08-29T17:22:12 <ThomasWaldmann> well, if you run one anyway and dislike the idea of having a gazillion files in the filesystem...
83 2010-08-29T17:22:48 <ThomasWaldmann> you can also do it on some different machine
84 2010-08-29T17:23:23 <jdwcornell> fair enough re: gazillion files. so that helps performance in some very large installations
85 2010-08-29T17:23:34 <ThomasWaldmann> maybe, maybe not
86 2010-08-29T17:24:04 <ThomasWaldmann> btw, the sqla backend needs much more love until it is really usable
87 2010-08-29T17:24:12 <ThomasWaldmann> dennda: ^^
88 2010-08-29T17:24:23 <jdwcornell> both rationales make sense. I try to avoid the "because I like it" rationales, preferring technical rationales :)
89 2010-08-29T17:24:54 <ThomasWaldmann> well, it often is just liking or disliking
90 2010-08-29T17:25:26 <ThomasWaldmann> especially since filesystems are rather effective these days also
91 2010-08-29T17:26:31 <jdwcornell> I can't tell you how many times I've been burnt by folks who make decisions based on "because I like it" ;) always try to get down to the technical / scientific / engineering rationale rather than working at the emotional level
92 2010-08-29T17:26:39 <jdwcornell> can't always do that though :)
93 2010-08-29T17:27:02 <jdwcornell> ... especially when trying to sell something ...
94 2010-08-29T17:30:22 <ThomasWaldmann> well, even on the technical level, there is no "one thing fits all" thing
95 2010-08-29T17:30:43 <jdwcornell> no, but there can be within design goals / parameters
96 2010-08-29T17:32:56 <ThomasWaldmann> one goal right now is to have something usable before hell freezes over :)
97 2010-08-29T17:33:18 <jdwcornell> I sense that it is a moin2 design goal to be flexible in backends, versus being the fastest rendering wiki
98 2010-08-29T17:33:21 <jdwcornell> LOL
99 2010-08-29T17:33:29 <jdwcornell> the tip works for me ;)
100 2010-08-29T17:33:57 <ThomasWaldmann> having the fastest rendering is not a primary goal I think
101 2010-08-29T17:34:59 <ThomasWaldmann> of course it shouldn't be slow either, but correctness, simplicity, flexibility is more important
102 2010-08-29T17:35:49 <ThomasWaldmann> btw, did you see the transclude stuff while it was working?
103 2010-08-29T17:36:02 <jdwcornell> and minimizing yak shearing :)
104 2010-08-29T17:36:04 <ThomasWaldmann> (it is broken since a week or two)
105 2010-08-29T17:36:07 <jdwcornell> no
106 2010-08-29T17:36:34 <ThomasWaldmann> that dom stuff makes it rather powerful
107 2010-08-29T17:37:02 <ThomasWaldmann> it is just hanging one dom tree into the other (at the point where the transclusion happens)
108 2010-08-29T17:37:11 <jdwcornell> I just jumped into this earlier today (dev wise). I'm doing my own yak shearing here... "I can deliver this to the client this way, but if I can solve this then I can deliver it to the client in less time" .... which usually ends up as greater wall-clock time even if I charge less client time :)
109 2010-08-29T17:37:33 <ThomasWaldmann> test.moinmo.in:8080 has a demo, but...
110 2010-08-29T17:38:50 <ThomasWaldmann> well, in case you want to use moin2 for production, be careful, it isn't ready nor stable for that
111 2010-08-29T17:39:26 <jdwcornell> pseudo-production. hard to describe this client relationship :)
112 2010-08-29T17:39:35 <jdwcornell> it is good enough for me now
113 2010-08-29T17:40:51 * ThomasWaldmann just thinks of amount of brokenness still and length of todo list...
114 2010-08-29T17:41:40 <ThomasWaldmann> http://test.moinmo.in:8080/+dom/MoinMoin2 seen that?
115 2010-08-29T17:42:11 <jdwcornell> not yet - was just checking out http://test.moinmo.in:8080/+get/5/transclude%20something
116 2010-08-29T17:43:43 <ThomasWaldmann> transclude is crashing rather often since 2w or so, needs debugging but developers busy otherwise
117 2010-08-29T17:47:15 <ThomasWaldmann> http://test.moinmo.in:8080/test.zip works now (click on the svg member)
118 2010-08-29T17:47:50 <jdwcornell> http://test.moinmo.in:8080/+dom/transclude%20something seems to too
119 2010-08-29T17:48:59 <jdwcornell> kudos for supporting machine readability
120 2010-08-29T17:49:11 <jdwcornell> that is, linked data / REST / and such
121 2010-08-29T17:49:38 <ThomasWaldmann> came by using flask framework :)
122 2010-08-29T17:51:10 <jdwcornell> :)
123 2010-08-29T17:53:58 <ThomasWaldmann> aside from hg stuff, do you have some special area of interest / expertise?
124 2010-08-29T17:54:20 <jdwcornell> heh that's a loaded question ;)
125 2010-08-29T17:54:42 <jdwcornell> lemme go get some more coffee and think about how to succinctly answer that
126 2010-08-29T17:59:28 <ThomasWaldmann> btw, we added jquery to moin static files, but only using it a little yet
127 2010-08-29T18:01:24 <ThomasWaldmann> i recently removed the openid auth from moin2 (mostly due to lack of maintainer)
128 2010-08-29T18:02:26 <jdwcornell> back.
129 2010-08-29T18:02:48 <jdwcornell> heh I'm a jQuery fanboy. even got my picture taken with Resig at the jquery conference last sept in Boston
130 2010-08-29T18:03:00 * ThomasWaldmann read your homepage :)
131 2010-08-29T18:03:14 <jdwcornell> ... and I use openid auth in my moinmoin instance
132 2010-08-29T18:03:27 <jdwcornell> ahhhh that makes those comments make sense :)
133 2010-08-29T18:04:24 <jdwcornell> soo.... what interests me right now? CMSes and Content Management, how to make it not suck
134 2010-08-29T18:05:00 <ThomasWaldmann> due to the big changes that have to be done to (old) moin code (and all over the place), I am sometimes removing functionality that isn't being maintained or actively worked on, if it doesn't qualify as "basic functionality" (which is release target of first moin2 release)
135 2010-08-29T18:05:01 <jdwcornell> I'm very interested in how people collaborate. for instance, how do you record those interactions so outside observers can become integrated into projects
136 2010-08-29T18:05:29 <jdwcornell> I've also been in management for a few years now
137 2010-08-29T18:05:57 <jdwcornell> I'm *huge* into linked data (semweb: rdf, owl, etc)
138 2010-08-29T18:06:20 <ThomasWaldmann> maybe you could help is with metadata and jquery
139 2010-08-29T18:06:26 <jdwcornell> but I've been finding it hard to make time these past few years to stay involved in any of it in anything more than the periphery
140 2010-08-29T18:07:04 <jdwcornell> and I'm trying to change that. Asking about the direction of parts of moin are the first step in that direction
141 2010-08-29T18:07:14 <ThomasWaldmann> i sometimes felt that our metadata key/values should follow some standard, but then I either found some mismatch or got confused :)
142 2010-08-29T18:08:34 <jdwcornell> I'm very *interested* in helping. First need to get acquainted and then see what my time allows, keeping in mind that I'm trying to *make* time. Just want to set expectations and don't want to be the person who talks a lot and never does anything
143 2010-08-29T18:09:09 <jdwcornell> ... I've been that guy, and don't want to be it again. And as author/maintainer of freeamp back in the late 90s I had my share of folks who did that. annoying after a while...
144 2010-08-29T18:09:18 <xorAxAx> do you know momo, jdwcornell? :)
145 2010-08-29T18:09:36 <jdwcornell> so does that life story answer your "what are you interested in" question? :)
146 2010-08-29T18:09:51 <ThomasWaldmann> yeah, thanks :)
147 2010-08-29T18:11:43 <jdwcornell> xorAxAx: no? googled it and would like to go to the restaurant
148 2010-08-29T18:12:02 <xorAxAx> jdwcornell: its a novel :)
149 2010-08-29T18:12:09 <xorAxAx> i was reminded of it when you talked about making time :)
150 2010-08-29T18:12:47 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9781:40bdab05b3de 2.0-dev/MoinMoin/ (25 files in 17 dirs): remove request params for some stuff that doesn't need it any more
151 2010-08-29T18:13:12 <jdwcornell> xorAxAx: fascinating. I should put that on my queue.
152 2010-08-29T18:13:23 <jdwcornell> xorAxAx: also, http://www.urbandictionary.com/define.php?term=momo
153 2010-08-29T18:13:50 <xorAxAx> awww
154 2010-08-29T18:16:58 <xorAxAx> jdwcornell: looks like you would need to order the book from the uk if you dont find a second hand version
155 2010-08-29T18:54:09 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9782:5d5f47f01f1d 2.0-dev/MoinMoin/ (14 files in 10 dirs): is*Page -> is*Item, remove unneded request params, misc. cleanup
156 2010-08-29T19:12:12 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9783:9a3070c7b8c8 2.0-dev/ (8 files in 7 dirs): renamed page_*_regex(act) -> item_*_regex(act)
157 2010-08-29T19:16:07 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9784:79b785ead56f 2.0-dev/MoinMoin/wikiutil.py: remove unused/deprecated getHomePage
158 2010-08-29T19:37:19 *** RogerHaase
159 2010-08-29T21:17:01 * ThomasWaldmann simplifies interwiki map stuff
160 2010-08-29T21:33:06 *** kursor
161 2010-08-29T21:34:26 *** kursor
162 2010-08-29T21:42:35 *** kursor
163 2010-08-29T21:43:08 *** kursor
164 2010-08-29T21:53:36 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9785:1f1271fd9b9e 2.0-dev/MoinMoin/ (12 files in 8 dirs): (log message trimmed)
165 2010-08-29T21:53:36 <CIA-36> simplify interwiki map code, see below
166 2010-08-29T21:53:36 <CIA-36> * remove unneeded request params
167 2010-08-29T21:53:36 <CIA-36> * removed loading interwiki map from a wiki page "InterWikiMap" feature
168 2010-08-29T21:53:36 <CIA-36> * removed loading interwiki map from a (list of) file(s) feature, removed
169 2010-08-29T21:53:37 <CIA-36> cfg.shared_intermap configuration attribute that was used to configure this
170 2010-08-29T21:53:38 <CIA-36> * added cfg.interwiki_map configuration attribute (dict wiki_name -> wiki_url)
171 2010-08-29T22:22:08 <dreimark> moin
172 2010-08-29T22:24:14 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9786:eaacd77d7314 2.0-dev/MoinMoin/ (theme/__init__.py user.py wikiutil.py): getInterwikiHomePage - remove unneeded request param
173 2010-08-29T22:24:17 <ThomasWaldmann> hi dreimark
174 2010-08-29T22:25:32 <dreimark> much to read in the backlog :)
175 2010-08-29T22:25:54 * dreimark wished we have already 2.0 have to do a new action with forms
176 2010-08-29T22:46:19 <ThomasWaldmann> you can use flatland, then it is less work (now and later)
177 2010-08-29T22:49:09 <dreimark> yes, with flatland we can have dynamic forms if that is needed too.
178 2010-08-29T23:12:40 *** grzywacz
179 2010-08-29T23:43:17 <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9787:2074a3c8c919 2.0-dev/MoinMoin/ (8 files in 6 dirs):
180 2010-08-29T23:43:17 <CIA-36> theme: removed unneeded request params, modify emit_custom_html
181 2010-08-29T23:43:17 <CIA-36> theme.make_icon: add formatter as first param
182 2010-08-29T23:43:17 <CIA-36> emit_custom_html: do not give request parameter to html callables
183 2010-08-29T23:43:17 <CIA-36> fix test_navi_bar theme tests
