1 2011-09-18T06:42:31 *** Marchael1
2 2011-09-18T08:32:32 *** Marchael1
3 2011-09-18T08:39:25 *** MattMaker
4 2011-09-18T08:43:47 *** marchael
5 2011-09-18T08:44:40 *** MattMaker
6 2011-09-18T08:48:12 *** marchael
7 2011-09-18T08:48:37 *** marchael
8 2011-09-18T08:52:06 *** MattMaker
9 2011-09-18T08:59:07 *** MattMaker
10 2011-09-18T09:22:35 *** raignarok
11 2011-09-18T09:36:47 <ronny> moin
12 2011-09-18T10:04:22 *** raignarok
13 2011-09-18T11:55:45 *** greg_f
14 2011-09-18T12:17:17 <ThomasWaldmann> moin
15 2011-09-18T13:31:58 * ThomasWaldmann cleans up a bit in storage-ng
16 2011-09-18T13:41:49 <ronny> ThomasWaldmann: im not sure what to do with acls tho, i dont have experience in permission management
17 2011-09-18T13:49:18 <ThomasWaldmann> we'll reuse acl computation as we have
18 2011-09-18T13:49:29 <ThomasWaldmann> only question is code structure
19 2011-09-18T13:49:42 <ThomasWaldmann> middleware layer, mixin, inline code
20 2011-09-18T14:15:42 * ThomasWaldmann pushed
21 2011-09-18T14:15:50 <ThomasWaldmann> cosmetic stuff
22 2011-09-18T14:17:40 <ronny> ThomasWaldmann: not sure what would work best
23 2011-09-18T14:18:13 <ThomasWaldmann> about package structure
24 2011-09-18T14:18:30 <ThomasWaldmann> in moin2 we currently have "storage" and everything related is below that
25 2011-09-18T14:18:43 <ThomasWaldmann> how about rearranging storage-ng that way first?
26 2011-09-18T14:18:57 <ThomasWaldmann> storage
27 2011-09-18T14:19:28 <ThomasWaldmann> storage.backends, storage.stores, storage.middleware?
28 2011-09-18T14:21:28 <ThomasWaldmann> or even put the middleware modules into storage package?
29 2011-09-18T14:24:14 <ThomasWaldmann> ronny:
30 2011-09-18T14:27:47 <ronny> ThomasWaldmann: how about rearranging it so, we can just delete moinmoin/storrage and then merge the new repo?
31 2011-09-18T14:29:23 <ThomasWaldmann> well, it will need more work afterwards anyway
32 2011-09-18T14:29:33 <ThomasWaldmann> but yes, what I said is part of that
33 2011-09-18T14:31:42 <ThomasWaldmann> maybe a separate branch for that will make sense, to not break the default branch
34 2011-09-18T14:32:18 <ThomasWaldmann> and also to keep the default branch for data export for some guys using moin2 in production :)
35 2011-09-18T14:33:03 <ronny> ThomasWaldmann: just taking a look at the acl middleware
36 2011-09-18T14:33:25 <ronny> ThomasWaldmann: the moinmoin2 acl middleware seems pretty reasonable
37 2011-09-18T15:04:57 <ThomasWaldmann> i'll look after finishing shuffling stuff around
38 2011-09-18T15:06:43 <ThomasWaldmann> pushed
39 2011-09-18T15:08:27 <ThomasWaldmann> ronny: we also need to check search*/document? methods
40 2011-09-18T15:09:03 <ronny> ThomasWaldmann: true
41 2011-09-18T15:15:22 <ThomasWaldmann> and the old acl stuff always fetched acls from storage, not from index
42 2011-09-18T15:16:40 <ronny> ThomasWaldmann: ok, its just the pattern that looks reasonable
43 2011-09-18T15:25:39 <ThomasWaldmann> pytest.importorskip('..kc')
44 2011-09-18T15:25:49 <ThomasWaldmann> -> not supported :(
45 2011-09-18T15:41:52 * ThomasWaldmann moves more stuff around
46 2011-09-18T15:50:58 <ronny> ThomasWaldmann: wow, making relavive imports is kinda hard
47 2011-09-18T15:54:34 <ronny> http://paste.pocoo.org/show/477966
48 2011-09-18T15:58:22 <ThomasWaldmann> pep8 fail in 2 :)
49 2011-09-18T15:58:43 <ThomasWaldmann> 5
50 2011-09-18T15:59:06 <ThomasWaldmann> 19
51 2011-09-18T15:59:40 <ronny> please done pep8.py on me
52 2011-09-18T15:59:52 * ThomasWaldmann haz a builtin pep8 checker
53 2011-09-18T16:01:09 <ThomasWaldmann> do you think it makes sense to put most of README.txt into storage.__doc__ ?
54 2011-09-18T16:02:01 <ronny> ThomasWaldmann: yeah, some can go to the subpackages/modules
55 2011-09-18T16:03:02 * ThomasWaldmann does that
56 2011-09-18T16:43:13 * ThomasWaldmann pushed
57 2011-09-18T17:23:24 * ThomasWaldmann rewrites create_simple_mapping
58 2011-09-18T17:50:37 *** rivanvx
59 2011-09-18T17:53:46 *** rivanvx
60 2011-09-18T17:56:35 *** rivanvx
61 2011-09-18T18:47:08 <ThomasWaldmann> ronny: http://paste.pocoo.org/show/478060/ something like that
62 2011-09-18T18:51:56 *** greg_f
63 2011-09-18T19:18:31 *** grzywacz
64 2011-09-18T19:18:31 *** grzywacz
65 2011-09-18T19:28:56 * ThomasWaldmann moves the stuff relevant to stores backend to Stores.from_url() classmethod
66 2011-09-18T19:58:19 <ronny> ThomasWaldmann: we should make a constants module with all the constant variables
67 2011-09-18T19:58:27 <ronny> ThomasWaldmann: and import it as C
68 2011-09-18T19:58:48 <ronny> (random thought i had)
69 2011-09-18T19:59:05 <ThomasWaldmann> for MTIME, SIZE etc?
70 2011-09-18T19:59:39 * ThomasWaldmann turns on the TV :)
71 2011-09-18T20:03:50 <ThomasWaldmann> \o/ Pirates! :D
72 2011-09-18T20:07:12 <ronny> ThomasWaldmann: yeah
73 2011-09-18T20:14:45 <dreimark> :) :)
74 2011-09-18T21:13:55 <ThomasWaldmann> ronny: http://paste.pocoo.org/show/478132/
75 2011-09-18T21:26:15 <ronny> ThomasWaldmann: looks like a good idea, partially broken yet
76 2011-09-18T21:28:40 <ronny> bbl, atm im watching a docu, and the player doesnt support pause
77 2011-09-18T21:29:08 <ThomasWaldmann> ronny: http://paste.pocoo.org/show/478143/ current state
78 2011-09-18T21:39:02 <ronny> ThomasWaldmann: wrt create_simple_mapping
79 2011-09-18T21:39:15 <ronny> the last few lines
80 2011-09-18T21:40:16 <ronny> mountpoints = sorted(namespace_mapping, key=len, reverse=True)
81 2011-09-18T21:40:59 <ronny> (seems more nice than the alternative to me
82 2011-09-18T21:41:38 <ThomasWaldmann> why key=len? i want len of the first tuple element.
83 2011-09-18T21:41:48 <ronny> ThomasWaldmann: no items there
84 2011-09-18T21:41:51 <ThomasWaldmann> not of the whole tuple, that is always 3
85 2011-09-18T21:42:29 <ThomasWaldmann> you are looking at the latest pastebin?
86 2011-09-18T21:42:32 <ronny> maybe im overoptimizing in my mind
87 2011-09-18T21:42:55 <ronny> hmm
88 2011-09-18T21:43:01 <ronny> ok, the last one changed a bit
89 2011-09-18T21:43:35 <ronny> ok, now i get it
90 2011-09-18T21:43:54 <ronny> so there will always be at least 3 backend with different base acl's
91 2011-09-18T21:45:15 <ronny> ThomasWaldmann: fromuri is messed wrt metastores vs bytestores
92 2011-09-18T21:50:20 <ThomasWaldmann> http://paste.pocoo.org/show/478157/ new iteration
93 2011-09-18T21:52:54 <ThomasWaldmann> ronny: i didn't understand what you mean is messed
94 2011-09-18T21:53:29 <ronny> ThomasWaldmann: something bout its doesnt sit well with me, i'll need a few minutes to get it into my concious mind
95 2011-09-18T21:55:42 <ronny> ThomasWaldmann: sqlite should extract the table name from the ui
96 2011-09-18T21:55:44 <ronny> uri
97 2011-09-18T21:58:57 <ThomasWaldmann> ronny: yes, such stuff can be added later
98 2011-09-18T21:59:07 <ThomasWaldmann> (for all stores, all backends)
99 2011-09-18T22:02:07 <ThomasWaldmann> http://paste.pocoo.org/show/478164/ fixed module loading
100 2011-09-18T22:09:34 <ronny> ThomasWaldmann: wouldnt it make sense to use newstyle format strings?
101 2011-09-18T22:10:14 <ronny> ThomasWaldmann: newstyle format strings are a lot more pleasant to read/write
102 2011-09-18T22:10:52 <ronny> ThomasWaldmann: we might want to be able to pass stuff like kind into from_url
103 2011-09-18T22:14:19 <ThomasWaldmann> we can change later to newstyle, the current stuff in moin2 does it quite similar to what i implemented
104 2011-09-18T22:15:58 <ronny> hmk
105 2011-09-18T22:16:27 <ThomasWaldmann> ok, commit?
106 2011-09-18T22:17:11 <ronny> i suppose, it needs some more testing, so backends constructed from uri expose the same behaviour as normal ones
107 2011-09-18T22:20:57 <ThomasWaldmann> i just noticed that this ns_mapping stuff is for old acl middleware
108 2011-09-18T22:32:39 <ThomasWaldmann> we also have a structural change
109 2011-09-18T22:32:57 <ThomasWaldmann> old: router -> acl -> backend
110 2011-09-18T22:33:28 <ThomasWaldmann> new: indexing/acl -> router -> backend
111 2011-09-18T22:35:06 <ThomasWaldmann> or alternative new: acl -> indexing -> router -> backend
112 2011-09-18T22:35:43 <ronny> ThomasWaldmann: so there is still some base setup for whats the base inherrited acl for stuff like mountpoints
113 2011-09-18T22:37:27 <ThomasWaldmann> yes, i guess it means we have acls for parts of namespace now (not for backends as in old)
114 2011-09-18T22:37:38 *** raignarok
115 2011-09-18T22:44:03 <ThomasWaldmann> ronny: btw, wrapping Indexer.search with a acl layer currently does not work good
116 2011-09-18T22:44:20 <ThomasWaldmann> it yields whoosh docs
117 2011-09-18T22:44:58 <ThomasWaldmann> maybe it should give that doc to Item also and yield the item
118 2011-09-18T22:47:58 <ronny> ThomasWaldmann: search yielding items sounds like a good idea
119 2011-09-18T22:48:12 <ronny> ThomasWaldmann: i wonder if for normal operations whoosh can be a implementation detail
120 2011-09-18T22:49:57 <ThomasWaldmann> hmm, i guess I'll try the acl middleware to get stuff cleanly separated
121 2011-09-18T22:50:27 * ThomasWaldmann committed/pushed the current state
122 2011-09-18T22:50:44 <ronny> ThomasWaldmann: so acl middleware wrapping around storage
123 2011-09-18T22:52:03 <ThomasWaldmann> hmm, in fact search is finding Revisions, not Items
124 2011-09-18T22:52:39 <ThomasWaldmann> so maybe we need more clear Revision first
125 2011-09-18T22:52:46 <ThomasWaldmann> ehrm, clever*
126 2011-09-18T22:53:14 <ThomasWaldmann> == something that does not access backend without need
127 2011-09-18T22:56:19 *** raignarok
128 2011-09-18T22:59:39 <ronny> ThomasWaldmann: if we make data for revisions optional, we might be able to have some nice things
129 2011-09-18T23:00:00 <ronny> (like items that act as directory listing based on metadata + search)
130 2011-09-18T23:00:24 <ronny> (also data can be lazy loaded then)
131 2011-09-18T23:04:39 <ronny> ThomasWaldmann: any oppinion on the constants.py idea?
132 2011-09-18T23:10:42 <ThomasWaldmann> yes, constants is fine. currently it is called config, but we can rename that then
133 2011-09-18T23:11:06 <ThomasWaldmann> http://paste.pocoo.org/show/478211/ how about that?
134 2011-09-18T23:11:15 <ronny> ThomasWaldmann: in moinmoin Config has another purpose
135 2011-09-18T23:11:30 <ronny> ThomasWaldmann: to me it makes sense to prepare the new names now
136 2011-09-18T23:11:44 <ronny> and then just merge with a moin2 named branch
137 2011-09-18T23:12:41 <ronny> ThomasWaldmann: btw, any oppinion on the notion of having revisions with more than one data items and revisions with no data item?
138 2011-09-18T23:13:37 <ThomasWaldmann> no data: no problem (that is just a small optimization from data=='')
139 2011-09-18T23:14:17 <ThomasWaldmann> but multipe datas like add a bit of complexity
140 2011-09-18T23:14:23 <ThomasWaldmann> likely*
141 2011-09-18T23:14:51 *** grzywacz
142 2011-09-18T23:15:12 <ronny> ThomasWaldmann: just wondering if i could have a whole directory tree as data
143 2011-09-18T23:15:40 <ronny> (however with using the same file store data for the same file)
144 2011-09-18T23:15:45 <ThomasWaldmann> (about the fileserver stuff: i noticed that it the directory stuff is not great yet, it should not simulate a wiki markup item)
145 2011-09-18T23:17:27 <ronny> should be a listing item, and use search
146 2011-09-18T23:17:32 <ThomasWaldmann> yup
147 2011-09-18T23:22:43 <ThomasWaldmann> hm, could also create a simple list of names (that is used as base for rendering, saves some index lookups)
148 2011-09-18T23:27:28 <ThomasWaldmann> http://paste.pocoo.org/show/478224/ fixed some issues
149 2011-09-18T23:27:38 *** rivanvx
150 2011-09-18T23:27:44 *** rivanvx
151 2011-09-18T23:28:37 *** rivanvx
152 2011-09-18T23:29:42 <ronny> ThomasWaldmann: i wouldnt call the class meta
153 2011-09-18T23:29:51 <ronny> its kind of a standin
154 2011-09-18T23:30:23 <ThomasWaldmann> it's metadata :)
155 2011-09-18T23:36:31 <ronny> n8\
156 2011-09-18T23:37:12 <ThomasWaldmann> gn ronny
157 2011-09-18T23:42:21 *** eSyr
158 2011-09-18T23:45:09 *** eSyr
159
