1 2011-05-01T01:19:02 *** MattMaker
2 2011-05-01T01:21:32 *** fayaz_
3 2011-05-01T02:21:15 *** PawelPacana
4 2011-05-01T05:37:04 *** fayaz_
5 2011-05-01T07:06:45 *** MattMaker
6 2011-05-01T11:35:11 *** MattMaker
7 2011-05-01T13:44:25 <ThomasWaldmann> moin
8 2011-05-01T14:34:40 * ThomasWaldmann misses some student activity
9 2011-05-01T15:07:16 *** greg_f
10 2011-05-01T16:05:53 *** jek
11 2011-05-01T16:30:59 <dreimark> moin
12 2011-05-01T16:42:11 <dreimark> ThomasWaldmann: http://paste.pocoo.org/show/381065/
13 2011-05-01T16:42:58 <dreimark> i played with that change ff3.0.16
14 2011-05-01T16:43:05 <dreimark> seems to work now.
15 2011-05-01T16:43:35 <dreimark> can you change that on the static server ?
16 2011-05-01T16:45:17 <dreimark> ThomasWaldmann: can that "upload" be used to upload a directory too?
17 2011-05-01T16:46:06 <dreimark> result could be an item hierarchy then
18 2011-05-01T17:15:20 * dreimark looks at ceebox
19 2011-05-01T17:32:14 *** RogerHaase
20 2011-05-01T17:36:02 <dreimark> RogerHaase: do you know http://catcubed.com/2008/12/23/ceebox-a-thickboxvideobox-mashup/
21 2011-05-01T17:38:09 <dreimark> http://plugins.jquery.com/project/ceebox
22 2011-05-01T17:44:29 <RogerHaase> dreimark: no, I have not heard of ceebox before . Looks interesting, and not only for images. I have been looking for a way to embed google search results into a wiki page instead of displaying the google site.
23 2011-05-01T17:44:48 <dreimark> it is MIT license
24 2011-05-01T17:45:13 <dreimark> i think we should have something which is able to show content of items
25 2011-05-01T17:45:21 <dreimark> not only as a DL link
26 2011-05-01T17:45:54 * dreimark waits for sinha ;)
27 2011-05-01T17:46:18 *** fayaz
28 2011-05-01T17:49:39 <dreimark> http://catcubed.com/2008/12/23/ceebox-a-thickboxvideobox-mashup/
29 2011-05-01T17:49:51 <dreimark> sinha ^
30 2011-05-01T17:50:26 <fayaz> hi,
31 2011-05-01T17:50:38 <fayaz> trying to run moin from 2.0-dev
32 2011-05-01T17:50:53 <fayaz> but stuck at make pagepacks...
33 2011-05-01T17:50:59 <fayaz> no such target
34 2011-05-01T17:52:23 <dreimark> you should run it from moin-2.0
35 2011-05-01T17:52:38 <dreimark> fayaz: 2.0-dev is outdated
36 2011-05-01T17:52:55 <dreimark> fayaz: http://hg.moinmo.in/moin/2.0
37 2011-05-01T17:53:07 <dreimark> http://static.moinmo.in/m2/html/admin/install.html
38 2011-05-01T17:53:28 <fayaz> dreimark: okay...
39 2011-05-01T17:53:39 <fayaz> thanks... i'll try it
40 2011-05-01T17:56:41 <dreimark> fayaz: what is your field of interests?
41 2011-05-01T17:57:02 <fayaz> python, qt, sqlite
42 2011-05-01T17:57:07 <fayaz> in that order...
43 2011-05-01T17:57:23 <fayaz> oh and now appengine too
44 2011-05-01T17:57:30 <dreimark> hehe
45 2011-05-01T17:57:55 <fayaz> i wanna help out in foss projects...
46 2011-05-01T17:58:08 <dreimark> http://test.moinmo.in/Home is rather current but not tip
47 2011-05-01T18:15:32 <RogerHaase> ThomasWaldmann: adding a new patch set seemed so hard last night, but this morning it is so easy: http://codereview.appspot.com/4432084/
48 2011-05-01T18:28:02 * dreimark added some comments
49 2011-05-01T18:28:52 * dreimark understand now what the comments is used for, sorry for my style
50 2011-05-01T18:29:00 <dreimark> lines
51 2011-05-01T18:49:32 <dreimark> RogerHaase: http://odyniec.net/projects/imgareaselect/
52 2011-05-01T18:49:50 <dreimark> another MIT/GPL tool
53 2011-05-01T18:50:31 <dreimark> for cropping
54 2011-05-01T18:53:57 *** fayaz
55 2011-05-01T18:55:30 <dreimark> hmm, can we use meta for storing all of this paramters or do we need an itemtype which is similiar to the container type but saves the state of the used functions
56 2011-05-01T18:57:09 <dreimark> e.g. if svg-edit is used to draw on an item, should it save on that item or create a new one
57 2011-05-01T18:57:56 *** RogerHaase
58 2011-05-01T19:00:28 *** RogerHaase
59 2011-05-01T19:06:52 <RogerHaase> dreimark: thanks for code review. This is a really nice tool. :-)
60 2011-05-01T19:07:36 <dreimark> RogerHaase: andi was showing it at the python bar camp two weeks ago
61 2011-05-01T19:08:27 <dreimark> i think it is better than our pastebin workflow and it helps to improve to make good changesets
62 2011-05-01T19:10:27 <dreimark> also it is much eaier to see what happens instead of searching irc logds
63 2011-05-01T19:10:31 <dreimark> -d
64 2011-05-01T19:11:21 * dreimark is sorry for the comments line, i thought it would be the mail subject
65 2011-05-01T19:11:56 <dreimark> and I am still tired from travelling home
66 2011-05-01T19:20:51 <RogerHaase> dreimark: I was also confused by changeset comment and email contents and got them reversed.
67 2011-05-01T19:21:55 <RogerHaase> dreimark: thanks for imareaselect tool. I can use that in my digitalrockart application. Keep them coming :-)
68 2011-05-01T19:23:06 <dreimark> :)
69 2011-05-01T19:43:30 *** greg_f
70 2011-05-01T20:19:04 *** jek
71 2011-05-01T20:23:05 *** MattMaker
72 2011-05-01T20:25:28 *** ronny
73 2011-05-01T20:27:36 *** ronny
74 2011-05-01T20:39:18 <ThomasWaldmann> re
75 2011-05-01T20:57:41 <ThomasWaldmann> RogerHaase: did you check whether the interwiki regex and the url regex work correctly?
76 2011-05-01T21:01:23 <ThomasWaldmann> dreimark: wb ;)
77 2011-05-01T21:13:05 <RogerHaase> ThomasWaldmann: yes. Does something look wrong?
78 2011-05-01T21:22:21 <ThomasWaldmann> how does it keep them apart?
79 2011-05-01T21:23:27 <ThomasWaldmann> RogerHaase:
80 2011-05-01T21:33:28 <RogerHaase> ThomasWaldmann: I am not sure I understand your question: if it looks like interwiki (has a : in middle), then it checks for valid interwiki site name. If not, it reconstructs link_url from site name and item name and continues with link processing
81 2011-05-01T21:52:56 <ThomasWaldmann> MoinMoin:foo/bar
82 2011-05-01T21:53:07 <ThomasWaldmann> http://foo/bar
83 2011-05-01T21:53:24 <ThomasWaldmann> mailto:foo@bar
84 2011-05-01T21:54:14 <ThomasWaldmann> you see, they are quite similar. and you can't even use the // to keep them apart.
85 2011-05-01T22:02:28 <RogerHaase> The first 2 work, mailto fails. mailto fails on http://test.moinmo.in/MailTo as well...
86 2011-05-01T22:02:40 <dreimark> RogerHaase: it must work with all url_schemas
87 2011-05-01T22:03:16 <dreimark> http://hg.moinmo.in/moin/2.0/file/321898d8273f/MoinMoin/config/__init__.py#l53
88 2011-05-01T22:03:17 <ThomasWaldmann> one of my comments on rietveld was to use "known" schemes as already done in the moinwiki19 code
89 2011-05-01T22:03:30 <ThomasWaldmann> (you can copy it from there)
90 2011-05-01T22:07:38 <RogerHaase> oh, missed the comment. I was only working on interwiki.
91 2011-05-01T22:10:31 <ThomasWaldmann> well, it's related. the regexes must be so always the right one matches.
92 2011-05-01T22:14:10 <ThomasWaldmann> (strictly taken, a URL has nothing one could use to recognize it, but it is good enough if we recognize it via the well known protocol schemes)
93 2011-05-01T22:16:52 <dreimark> ThomasWaldmann: http://www.web-delicious.com/jquery-plugins-demo/wdContextMenu/sample.htm
94 2011-05-01T22:17:11 <dreimark> what do you think on adding a right click menu to an item view
95 2011-05-01T22:17:39 <dreimark> with the common manipulation tools?
96 2011-05-01T22:18:15 <ThomasWaldmann> what's the problem you are trying to solve?
97 2011-05-01T22:19:38 <dreimark> move "Copy, Rename, Delete, Destroy" from the edit bar
98 2011-05-01T22:20:03 <dreimark> and some more
99 2011-05-01T22:20:32 <ThomasWaldmann> that solves what?
100 2011-05-01T22:20:51 <dreimark> the long list of commands
101 2011-05-01T22:21:45 <ThomasWaldmann> well, you still have them, just at another place. we had them in a menu until recently.
102 2011-05-01T22:21:56 <dreimark> it can be used to speparate item related commands from global ones
103 2011-05-01T22:22:50 <dreimark> we don't use right mouse click, the question is more if we want
104 2011-05-01T22:23:26 <ThomasWaldmann> a right click usually shows a browser menu
105 2011-05-01T22:25:42 <ThomasWaldmann> i am in favour if making some senseful split between global actions and item-local actions
106 2011-05-01T22:27:13 <ThomasWaldmann> what arrangement or technology we use for this has to be seen. usually users don't right-click anywhere when surfing the web. if you don't have a 2button mouse, but a 1button or a touchpad or ... you maybe can't easily right-click.
107 2011-05-01T22:28:54 <dreimark> may be we should also look for multi touch features
108 2011-05-01T22:29:20 <ThomasWaldmann> maybe we should get the basic simple stuff working first? :)
109 2011-05-01T22:29:23 <dreimark> right mouse click is used in all non browser applications
110 2011-05-01T22:29:57 <dreimark> ThomasWaldmann: yes, but we should not miss to look for inspirations too
111 2011-05-01T22:31:00 <ThomasWaldmann> sure. but don't use unusual and unexpected stuff if it can be solved in a usual way.
112 2011-05-01T22:32:57 <ThomasWaldmann> (you see it on that demo site, btw. the button text says "right click here". why? because if it would not be printed on the button, noone would ever right-click there.)
113 2011-05-01T22:35:15 <dreimark> http://tikku.com/scripts/ui/radmenu/radmenu_example.html#
114 2011-05-01T22:36:01 <dreimark> mostly user "know" that right click is not usefull with browser apps
115 2011-05-01T22:36:25 <dreimark> this is the reason why they forget the right button
116 2011-05-01T22:36:58 <ThomasWaldmann> yeah, and this is one reason why one should not use that
117 2011-05-01T22:37:02 <dreimark> new users try always the right button and are surprised that it has no context for the app
118 2011-05-01T22:37:35 <dreimark> the rad menue is funny
119 2011-05-01T22:41:20 <ThomasWaldmann> maybe we can help all together on first improving what we have
120 2011-05-01T22:41:49 <ThomasWaldmann> like e.g. the upload stuff, it still needs css fixes and good integration/placement
121 2011-05-01T22:43:12 * ThomasWaldmann imagines that this should be in a sidebar or in a footer area, that maybe is collapsed by default and filled on expanding it.
122 2011-05-01T22:44:19 <dreimark> ThomasWaldmann: is it indented that it could only used on an existing item?
123 2011-05-01T22:44:28 <dreimark> or should that change too?
124 2011-05-01T22:44:55 <ThomasWaldmann> the upload?
125 2011-05-01T22:45:01 <dreimark> yes
126 2011-05-01T22:45:19 <ThomasWaldmann> that is unrelated to current item existance
127 2011-05-01T22:45:30 <dreimark> http://test.moinmo.in/+index2/
128 2011-05-01T22:45:44 <ThomasWaldmann> it is allowed to up- or download subitems even if current item does not exist
129 2011-05-01T22:46:13 <dreimark> http://test.moinmo.in/+index2/dd
130 2011-05-01T22:46:20 <dreimark> makes then subitems of dd ?
131 2011-05-01T22:46:30 <dreimark> but not items of /
132 2011-05-01T22:47:10 <ThomasWaldmann> well, you can't get to /+index/ via the UI
133 2011-05-01T22:47:34 <ThomasWaldmann> we can think about whether it makes sense to use it to create multiple toplevel items
134 2011-05-01T22:48:07 <dreimark> i think we should not restrict this
135 2011-05-01T22:48:43 <dreimark> you can populate a wiki by this function easily
136 2011-05-01T22:50:41 <dreimark> an other aspect / feature is to add a whole directory
137 2011-05-01T22:51:34 <ThomasWaldmann> you can do that, use ctrl-a
138 2011-05-01T22:52:15 <ThomasWaldmann> the problem with that easy multi-file upload is that we don't have too much metadata that way
139 2011-05-01T22:52:28 <ThomasWaldmann> (at least in the current implementation)
140 2011-05-01T22:52:31 <ThomasWaldmann> no comment
141 2011-05-01T22:52:45 <ThomasWaldmann> mimetype depends on filename extension
142 2011-05-01T22:53:01 <dreimark> yes
143 2011-05-01T22:53:50 <ThomasWaldmann> if the extension is wrong, revision will get created with incorrect metadata
144 2011-05-01T22:55:48 <dreimark> or as "application/octet-stream" if there is no extension
145 2011-05-01T22:56:07 <ThomasWaldmann> same thing
146 2011-05-01T22:56:34 <dreimark> i was meaning more like adding my Desktop directory and it becomes expaded into wiki items
147 2011-05-01T22:56:52 <dreimark> but that is not important
148 2011-05-01T22:57:35 <ThomasWaldmann> the missing comment can be fixed quite easily, but the per file metadata is more difficult
149 2011-05-01T22:58:01 <ThomasWaldmann> (you also can't set an acl currently)
150 2011-05-01T22:58:05 <dreimark> yes, the only tool i know which give the exact answer is unix file
151 2011-05-01T22:58:15 <ThomasWaldmann> not even that
152 2011-05-01T22:58:50 <ThomasWaldmann> it's the same as with encoding. you can try guessing, but in the end, you have to *know*.
153 2011-05-01T23:00:17 <dreimark> yes, it will only find that it is e.g. ascii which means major type text/ but it won't be able to find the minor type
154 2011-05-01T23:00:33 <dreimark> and it is impossible to find the x-* types
155 2011-05-01T23:01:26 <ThomasWaldmann> what we could maybe do (not sure how), is to guess mimetype for each files and display the guessed type to the user, with the possibility to fix it, and then transfer all info on upload somehow
156 2011-05-01T23:01:27 <dreimark> additional to highlight we could have a plain action
157 2011-05-01T23:01:56 <dreimark> which was raw in the past
158 2011-05-01T23:03:12 <dreimark> we better don't ask the user that will end in a gau
159 2011-05-01T23:04:26 <dreimark> it would work may be the other way if we ask the user to select the mimetype he wants to upload
160 2011-05-01T23:04:27 <ThomasWaldmann> well, if you do automatic guessing first and offer a way to correct it, it should be better than just autodetect
161 2011-05-01T23:04:32 <dreimark> and filter his list by that
162 2011-05-01T23:04:37 <ThomasWaldmann> (assuming users are not idiots :)
163 2011-05-01T23:05:18 <dreimark> that would let the user one type after the other
164 2011-05-01T23:05:31 <dreimark> +upload
165 2011-05-01T23:05:53 <ThomasWaldmann> that would either degrade the multifile-upload to single file in the worst case (if you could select only one mimetype for all)
166 2011-05-01T23:06:10 <ThomasWaldmann> or it would be more complicated than what i said
167 2011-05-01T23:06:41 <dreimark> or we allow it only for extensions we know
168 2011-05-01T23:06:54 <dreimark> and ask for feature requests for unknown
169 2011-05-01T23:07:51 <ThomasWaldmann> that won't work
170 2011-05-01T23:08:28 <dreimark> we can have a comfortable upload for these
171 2011-05-01T23:08:35 <dreimark> and a one by on for others
172 2011-05-01T23:08:36 <ThomasWaldmann> the extension / mimetype system is fscked up enough that it just won't work automatically
173 2011-05-01T23:09:07 <ThomasWaldmann> e.g. if you upload a bunch of *.dot, what is that? :)
174 2011-05-01T23:09:34 <ThomasWaldmann> or README and COPYING and OHWELL, what's that?
175 2011-05-01T23:10:22 <dreimark> if you have to answer a trillion of questions for uploading that stuff you start to hate the app
176 2011-05-01T23:10:45 <dreimark> i think we should use it only on a restricted set of mimetypes
177 2011-05-01T23:11:00 <dreimark> e.g. we have in our item list
178 2011-05-01T23:11:30 <dreimark> and if one wants to add another one, he should do it one by one
179 2011-05-01T23:11:44 <dreimark> or add some code so that it is known what it is
180 2011-05-01T23:13:05 <dreimark> and the ones without extension - i prefer them too. there is no idea by the user if it is text/plain or e.g. text/x-rest
181 2011-05-01T23:13:43 <dreimark> text/x-rst
182 2011-05-01T23:14:57 <ThomasWaldmann> ah and then there is charset for the textfiles
183 2011-05-01T23:15:14 <ThomasWaldmann> we can't know or detect that either
184 2011-05-01T23:16:03 <ThomasWaldmann> (ascii works, maybe for utf-8, but bad chances for other stuff)
185 2011-05-01T23:16:27 <dreimark> we need to know if it is ascii or utf-8 and reject all orhers
186 2011-05-01T23:17:11 <dreimark> the user can use iconv on his system if he knows that this is the reason why he is not able to upload it
187 2011-05-01T23:17:43 <ThomasWaldmann> well, but maybe conversion is not wanted
188 2011-05-01T23:18:18 <dreimark> it is at least if you work with a current system
189 2011-05-01T23:18:26 <ThomasWaldmann> we can represent any text file, we can just have text/plain;charset=iso-8859-1 as content-type in metadata
190 2011-05-01T23:18:46 <dreimark> because lots of editors don't save if they detect a charset mismatch
191 2011-05-01T23:19:16 <dreimark> i think we should not save anything else than utf-8
192 2011-05-01T23:20:19 <dreimark> if we save something else we have no chance to restrict someone to alter something with an other charset
193 2011-05-01T23:20:40 <ThomasWaldmann> ?
194 2011-05-01T23:20:43 <dreimark> and then it is mixed up and similiar broken to the cms we used in our company before
195 2011-05-01T23:21:20 <dreimark> there are lots of encodings having german umlauts on differnt locations
196 2011-05-01T23:21:45 <ThomasWaldmann> btw, as long as we don't know the type for sure (== knowing it is text), we can't reject non-utf8 text either
197 2011-05-01T23:22:20 <dreimark> somehow my latex editor kile knows it
198 2011-05-01T23:22:26 *** jek
199 2011-05-01T23:22:26 *** jek
200 2011-05-01T23:22:28 <ThomasWaldmann> so if someone uploads a LIESMICH in latin-1, we won't know
201 2011-05-01T23:24:49 <dreimark> it tells
202 2011-05-01T23:24:51 <dreimark> The selected encoding cannot encode every unicode character in this document. Do you really want to save it? There could be some data lost.
203 2011-05-01T23:25:10 <dreimark> we should do something similiar
204 2011-05-01T23:26:34 <ThomasWaldmann> heuristics
205 2011-05-01T23:27:35 <dreimark> show an example where such a test is wrong in the manner that we can't store it
206 2011-05-01T23:28:00 <ThomasWaldmann> i guess we can't solve that. we'll just accept everything as it comes and if it is wrong, you have to edit metadata.
207 2011-05-01T23:28:58 <ThomasWaldmann> dreimark: if you have a file like XY with only a few bytes inside, you won't be able to guess type, nor encoding.
208 2011-05-01T23:33:43 <ThomasWaldmann> and i have seen ridiculous fails of linux' "file" command
209 2011-05-01T23:38:20 <dreimark> hmm, we need to be able to edit meta data without beeing able to create a new data revision
210 2011-05-01T23:38:46 <dreimark> otherwise you could have manual corrections for a wrong encoding in the same revisions
211 2011-05-01T23:39:04 <ThomasWaldmann> no, that won't work with current storage
212 2011-05-01T23:39:16 <ThomasWaldmann> we always revision meta in parallel to data
213 2011-05-01T23:42:29 <dreimark> if there is an acl on the item it also protects meta while it is written in meta
214 2011-05-01T23:43:14 <ThomasWaldmann> so?
215 2011-05-01T23:43:14 <dreimark> can we have a right which did not allow to alter data?
216 2011-05-01T23:43:39 <ThomasWaldmann> -write?
217 2011-05-01T23:44:03 <dreimark> could then meta be changed ?
218 2011-05-01T23:44:30 <dreimark> i wonder currently how meta and acls work
219 2011-05-01T23:44:31 <ThomasWaldmann> no
220 2011-05-01T23:44:57 <dreimark> -write_data
221 2011-05-01T23:46:34 <RogerHaase> * worries that above discussion may become another part of interwiki scope creep
222 2011-05-01T23:47:01 <ThomasWaldmann> RogerHaase: :D
223 2011-05-01T23:48:02 <ThomasWaldmann> RogerHaase: you are lucky, a link doesn't care for target type or encoding
224 2011-05-01T23:48:24 <ThomasWaldmann> (but transclusion is different)
225 2011-05-01T23:52:15 <dreimark> one thing we can do is to block data for editing if we fear a conflict and force one to fix it
226 2011-05-01T23:52:36 <RogerHaase> [[mailto://foo@bar]] works, should it? And the list of url_schemas in config has many more types than we have icons for external links in common.css.
227 2011-05-01T23:53:21 <dreimark> and because we can only once emit the charset used on the page
228 2011-05-01T23:53:34 <dreimark> we need to deliver every transcluded item in utf-8
229 2011-05-01T23:54:55 <dreimark> .oO this becomes more and more a cms style workflow
230 2011-05-01T23:56:10 <dreimark> it looks like one has to review before publishing mass uploaded items
231 2011-05-01T23:56:28 <ThomasWaldmann> RogerHaase: mailto: doesn't have the //netloc part
232 2011-05-01T23:57:03 * dreimark looks for content services
233 2011-05-01T23:57:28 <ThomasWaldmann> (and I didn't expect that you also do the fix to use url_schemas right now, that could be in a separate change)