2008-06-24T00:23:27 <PawelPacana> johill: dennda: should create_item(existing_item_name) raise error immediately or fall only on commit try? i thought the second one, as return Item is just a proxy, however memory backend tests state the first one
2008-06-24T01:28:56 * zenhase is feeling a bit nauseous right now
2008-06-24T01:29:53 <zenhase> i just helped the spokesperson of hour student dorm to investigate a case of a missing student
2008-06-24T03:16:19 <CIA-52> Reimar Bauer <rb.proj AT googlemail DOT com> default * 98:edb3aa8208b2 1.7-extensions/data/plugin/parser/text_x_arnica.py: text_x_arnica: removed some whitespaces, moved _get_files below macro_arnica
2008-06-24T03:16:19 <CIA-52> Reimar Bauer <rb.proj AT googlemail DOT com> default * 99:4762bad57dae 1.7-extensions/ (2 files in 2 dirs): arnica_slides: initial version basic css definition
2008-06-24T08:12:16 <ThomasWaldmann> moin
2008-06-24T08:27:17 <PawelPacana> ThomasWaldmann: moin
2008-06-24T09:02:21 <dreimark> moin
2008-06-24T09:06:58 <dreimark> ThomasWaldmann: I do want to move some of the arnica code to some places on MM 1.8, anica_slides.send_cache_file, text_x_arnica.arnica_alias_text to wikiutil.
2008-06-24T09:07:33 <dreimark> s/text_x_arnica.arnica_alias_text/text_x_arnica.get_exif_info/
2008-06-24T09:30:19 <dreimark> bbl
2008-06-24T09:46:24 <dreimark> re
2008-06-24T09:48:20 <dreimark> johill: if you have some time can you please review arnica
2008-06-24T09:55:22 <xorAxAx> moin melita
2008-06-24T09:58:07 <dreimark> moin Alexander
2008-06-24T09:58:12 <dreimark> moin melita
2008-06-24T10:00:34 <xorAxAx> moin reimar
2008-06-24T10:02:33 <dreimark> do you stay longer in the us or do I meet you also at vilnius
2008-06-24T10:02:57 <ThomasWaldmann> dreimark: you are partly reimplementing the "caching" code instead of just using it
2008-06-24T10:04:07 <dreimark> ThomasWaldmann: then I did something wrong
2008-06-24T10:04:35 <ThomasWaldmann> and as you need a URL for showing images, that could be just an action (not wikiutil)
2008-06-24T10:05:52 <xorAxAx> dreimark: me? i will be in the US at exactly the EP days
2008-06-24T10:07:53 <dennda> PawelPacana: let me think about that
2008-06-24T10:08:01 <dreimark> xorAxAx: oh it was not clear to me that it completly overlaps
2008-06-24T10:08:53 * ThomasWaldmann .oO(worst case scheduling)
2008-06-24T10:10:08 <ThomasWaldmann> otoh, both events will be fun :)
2008-06-24T10:11:11 <ThomasWaldmann> xorAxAx: btw, where's federico? :)
2008-06-24T10:11:52 * dreimark remembers that I wanted to write some ghop notes
2008-06-24T10:12:51 <xorAxAx> ThomasWaldmann: on various mailing lists :)
2008-06-24T10:15:42 <ThomasWaldmann> dreimark: if the code using "sendcache" action stores pre-rendered stuff (e.g. images) to the caching system using a non-predictable acess key, there would be no need to check / handle ACLs in the action
2008-06-24T10:17:00 <ThomasWaldmann> e.g. sha(secret + itemname) or sha(secret + parsercontentarea)
2008-06-24T10:18:11 <dennda> PawelPacana: I will answer later, lecture right now
2008-06-24T10:18:17 <dreimark> you do want to have them direct served?
2008-06-24T10:18:48 <dreimark> I am not sure if this is needed, I was wondering how fast arnica is
2008-06-24T10:19:46 <ThomasWaldmann> not direct served (in the apache sense), but served without much checking / error handling
2008-06-24T10:20:59 <ThomasWaldmann> that action could be generally useful for all extensions that expensively render stuff
2008-06-24T10:21:21 <ThomasWaldmann> and need to store pre-rendered stuff accessible via an url
2008-06-24T10:22:02 * dreimark has this impression too had some discussion with Oliver who has send me some functions to make accessible thumbn<ils
2008-06-24T10:23:12 <ThomasWaldmann> a problem that should be solved somehow is how to cleanup such stuff
2008-06-24T10:23:45 <dreimark> main cleancache
2008-06-24T10:23:52 <dreimark> +t
2008-06-24T10:24:30 <ThomasWaldmann> X)
2008-06-24T10:25:49 <ThomasWaldmann> dreimark: the cache content should have everything that is needed to emit a raw object
2008-06-24T10:25:50 <dreimark> I do prefer a naming convention so that the thumbnail don't have to rebuilded all the time for a different plugin
2008-06-24T10:27:11 <dreimark> ThomasWaldmann: I think send_cach_file can currently emit every mimetype
2008-06-24T10:27:22 <ThomasWaldmann> i.e. content-type, content-length maybe, content-disposition, filename)
2008-06-24T10:27:49 <dreimark> is all there
2008-06-24T10:27:52 <ThomasWaldmann> it is just too complicated
2008-06-24T10:28:47 * ThomasWaldmann thinks about a function with very few statements
2008-06-24T10:30:10 <dennda> PawelPacana: I think you are right. Let's see what johill says
2008-06-24T10:30:23 <dennda> But I think I'll need to change that
2008-06-24T10:30:35 * dreimark has added the check for pictures this is not needed (wanted not to have served the other html cache files)
2008-06-24T10:31:45 <ThomasWaldmann> i.e. .... timestamp
2008-06-24T10:32:19 <PawelPacana> dennda: from my point of view it better fits to convention
2008-06-24T10:32:25 <ThomasWaldmann> the url would give all needed to open the cache
2008-06-24T10:33:02 <ThomasWaldmann> the cache would give all needed to make the http headers and also give the content
2008-06-24T10:34:06 <dreimark> yeah, I missed the point that we need to move away from os.path
2008-06-24T10:35:19 <ThomasWaldmann> btw, for effectively handling larger cached files, I guess we even need 2 cache objects
2008-06-24T10:35:54 <ThomasWaldmann> one with a dict with the header data (to handle 304 and header generation)
2008-06-24T10:37:50 <dreimark> and the other for data only?
2008-06-24T10:38:05 <ThomasWaldmann> a second with the raw file data (and file-like access to it to use sendfile)
2008-06-24T10:38:21 <dennda> PawelPacana: the latter?
2008-06-24T10:38:24 <ThomasWaldmann> so that we don't need to read the file into memory
2008-06-24T10:38:35 <dennda> i.e. raising the exception in commit_item?
2008-06-24T10:39:31 <ThomasWaldmann> dreimark: could be just <hash>.meta and <hash>.data
2008-06-24T10:40:30 <PawelPacana> dennda: yes, but wait a second
2008-06-24T10:43:23 <PawelPacana> dennda: i forgot that create_item causes has_item to return True
2008-06-24T10:43:49 <dreimark> ThomasWaldmann: the non-predictable key needs to be different for each attachment
2008-06-24T10:44:21 <PawelPacana> and with this assumption your implementation is right
2008-06-24T10:44:29 <dennda> PawelPacana: don't think about has_item -- that is just a little helper method for convenience and should not affect api design
2008-06-24T10:44:51 <ThomasWaldmann> e.g. sha(secret + itemname) or sha(secret + parsercontentarea)
2008-06-24T10:45:34 <ThomasWaldmann> (and itemname == pagename/attachname
2008-06-24T10:46:24 <ThomasWaldmann> and for the parser, the same parser type + same content gives the same output anyway)
2008-06-24T10:46:49 * dreimark is not sure if we can avoid an acl check for an subitem
2008-06-24T10:47:06 <dreimark> if one has a bookmark of the action
2008-06-24T10:48:13 * johill returns
2008-06-24T10:48:16 <ThomasWaldmann> you could also look into the browser's image cache :)
2008-06-24T10:48:20 <johill> loads of question :)
2008-06-24T10:48:33 <johill> ThomasWaldmann: what was that about not sorting?
2008-06-24T10:48:56 <dennda> welcome back :) we missed you already
2008-06-24T10:49:06 <dreimark> hi johill
2008-06-24T10:49:39 <ThomasWaldmann> johill: just look at the macro output now. some rarely used settings are at top of the table because they sort there.
2008-06-24T10:50:15 <ThomasWaldmann> thus, I think we should just arrange once and manually in the cfg
2008-06-24T10:51:23 <ThomasWaldmann> and the magic for stuff with same prefix is nice, but as we don't have completely clean setting names, it would also be better to do it manually
2008-06-24T10:51:58 <dreimark> bbl
2008-06-24T10:53:39 <johill> ThomasWaldmann: the magic with prefix is just used for those setting groups that allow it, hence two things, I think we should keep it and only add new settings to the group-name ones so they're clean in the future
2008-06-24T10:54:25 <johill> wrt. sorting, I don't know, I think they're easier to find that way, though I guess rarely used once ... maybe we should have a "Commonly used settings" and a "Other settings" group instead
2008-06-24T11:25:48 <ThomasWaldmann> johill: the problem is that that magic generates 2 separate sections, even for stuff that should be in one
2008-06-24T11:26:46 <ThomasWaldmann> e.g. we have a section about xapian (could be: a section about search) and a separate setting that is available for all search engine types (search_result_count per page)
2008-06-24T11:27:37 <ThomasWaldmann> also, we have a section about acls and also a section about other security stuff
2008-06-24T11:28:45 <johill> right
2008-06-24T11:28:52 <johill> so we should move them to the no-group part for now
2008-06-24T11:29:15 <johill> I still think we should try to have appropriate prefixes
2008-06-24T11:29:53 <johill> say we add all storage stuff in a "data" section
2008-06-24T11:30:04 <ThomasWaldmann> sure, but that is not possible short-term for everything and maybe not even possible in every case
2008-06-24T11:31:15 <johill> no, of course that's not possible
2008-06-24T11:31:17 <johill> short term
2008-06-24T11:31:28 <johill> but for entirely new things, I think we should do that
2008-06-24T11:32:04 * ThomasWaldmann has even put some comment advising that somewhere some time ago iirc
2008-06-24T11:32:04 <johill> say we merge the classification stuff, for all the new settings that introduces (not that I know if/how many) it would be possible
2008-06-24T11:32:11 <johill> yeah you did
2008-06-24T11:32:31 <johill> so imho we should keep the feature even if we now need to move things out of the magic group thing
2008-06-24T11:32:41 <johill> me moving that there was more to try it out
2008-06-24T11:36:08 <ThomasWaldmann> ok
2008-06-24T11:37:53 <johill> so yes, I agree that it's very possible we cannot use the grouping now
2008-06-24T11:38:01 <johill> but I don't think we should take out the possibility again
2008-06-24T11:43:07 * ThomasWaldmann implements sendcached action
2008-06-24T11:49:30 <ThomasWaldmann> fpletz: that xapian locking stuff is evil
2008-06-24T11:50:01 <ThomasWaldmann> when i try to rebuild the index for moinmo.in, it often tells me that it couldn't acquire the lock
2008-06-24T11:50:26 <ThomasWaldmann> (and I don't know if this is because someone is using the index currently or if it is just some malfunction)
2008-06-24T11:52:11 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3737:c62beb39c64c 1.7/MoinMoin/search/queryparser.py: queryparser: reject invalid queries consisting only of "OR"
2008-06-24T11:52:54 <ThomasWaldmann> what about "not only or"? :)
2008-06-24T11:53:04 <johill> not only or?
2008-06-24T11:53:14 <ThomasWaldmann> foo or
2008-06-24T11:53:21 <johill> oh, well that ignores the OR
2008-06-24T11:53:44 <ThomasWaldmann> or foo also?
2008-06-24T11:54:02 <johill> that, umm, I need to fix
2008-06-24T11:54:19 <ThomasWaldmann> or or or something else :)
2008-06-24T11:54:22 <zenhase> hmm ... foo or? msgs.append(_('... or else what?')) ;)
2008-06-24T11:54:39 <johill> heh
2008-06-24T11:54:42 <ThomasWaldmann> zenhase: :)
2008-06-24T11:55:08 <zenhase> the search-engine equivalent of a hollow threat
2008-06-24T11:55:27 <johill> now I'm confused
2008-06-24T11:56:05 <johill> oh no
2008-06-24T11:56:12 <johill> I'm testing against moinmo.in instead of localhost ;)
2008-06-24T11:56:32 <johill> or foo actually does give an error, while 'foo or' ignores the or
2008-06-24T11:57:40 <johill> ThomasWaldmann: "use non-guessable cache key as action url param, so we don't have to check ACLs"?
2008-06-24T11:57:58 <johill> how can that work, the key would have to be per-user or so?
2008-06-24T12:03:29 <johill> ThomasWaldmann: about editor_quickhelp... maybe that should just be imported from the parser itself?
2008-06-24T12:16:51 <dreimark> and the non-guessable cache key needs to be changed if one overwrites attachments
2008-06-24T12:17:27 * dreimark prefers checking for acls
2008-06-24T12:18:32 <dreimark> the browsers cache file is old and one has had access to it but probably not to the new one
2008-06-24T12:20:09 <johill> not chacking acls would be neat
2008-06-24T12:20:23 <johill> then we could ask lighty to serve the files directly if we know to be running under it
2008-06-24T12:20:47 <johill> actually we can check acls even then, never mind
2008-06-24T12:25:48 <dreimark> yeah if it works. acls are still a characteristic of moinmoin. circumventing them save must be really save
2008-06-24T12:26:28 <johill> well all you have to do is do X-lighty-sendfile: /path/to/file or something
2008-06-24T12:26:32 <johill> instead of serving it out
2008-06-24T12:26:35 <johill> you can do whatever you want before that
2008-06-24T12:26:42 <johill> but it is supposed to be a lot faster
2008-06-24T12:27:20 <dreimark> cool
2008-06-24T12:27:29 <johill> ThomasWaldmann: imho we should group things sematnically rather than technically
2008-06-24T12:27:41 <johill> ThomasWaldmann: hence, page_license_page should be with the other editor stuff
2008-06-24T12:28:12 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3783:d9431d180aff 1.8/MoinMoin/ (config/multiconfig.py macro/WikiConfig.py): config improvements wrt. computed settings
2008-06-24T12:31:20 <dreimark> twikidraw can save in the caching area too
2008-06-24T12:35:10 <dreimark> bbl
2008-06-24T12:39:09 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3784:951ed1df8ec3 1.8/MoinMoin/ (6 files in 4 dirs):
2008-06-24T12:39:09 <CIA-52> move editor quickhelp out of config into parser modules
2008-06-24T12:39:09 <CIA-52> that way, custom parser modules can ship their quickhelp instead of
2008-06-24T12:39:09 <CIA-52> requiring the admin who enables them to put the quickhelp into the
2008-06-24T12:39:09 <CIA-52> config and it cleans up the config too.
2008-06-24T12:47:21 <CIA-52> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3785:651f505bd9f1 1.8/MoinMoin/action/sendcached.py: sendcached action (completely untested)
2008-06-24T12:47:21 <CIA-52> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3786:17ec10ccdae2 1.8/MoinMoin/ (7 files in 5 dirs): merged main
2008-06-24T12:47:44 <ThomasWaldmann> dreimark: ^^^^ please write some code using it :)
2008-06-24T12:49:58 <johill> ThomasWaldmann: I don't think 'somethingnotguessable' works
2008-06-24T12:50:37 <johill> ThomasWaldmann: as soon as somebody has the URL, it's not longer 'not guessable'
2008-06-24T12:50:49 <johill> since it doesn't very by user
2008-06-24T12:51:22 <johill> s/very/vary/
2008-06-24T12:53:45 <johill> also, the 'generate cached key' should be easier than opening the sendcache cache arena stuff etc.
2008-06-24T12:54:37 <ThomasWaldmann> johill: how should somebody have the url?
2008-06-24T12:55:07 <ThomasWaldmann> if someone has access to browser history or browser cache of a trusted user, he has the content anyway
2008-06-24T12:56:06 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3787:f07963fce93a 1.8/docs/CHANGES: document editor_quickhelp change
2008-06-24T12:56:24 <johill> ThomasWaldmann: well for starters you can't revoke rights that way, very bad
2008-06-24T12:57:09 <ThomasWaldmann> (and of course varying it by user would pump the cache to N times size)
2008-06-24T12:57:41 <johill> well you don't have to store it
2008-06-24T12:57:52 <johill> you could store the keys and the metadata into the users session
2008-06-24T12:57:57 <johill> and the filename too
2008-06-24T12:58:20 <johill> so you wouldn't need a cache at all and the key could be randomly generated on the fly for that session
2008-06-24T12:59:13 <ThomasWaldmann> it has 2 caches currently, meta and data
2008-06-24T12:59:33 <ThomasWaldmann> (and you have to store the data somewhere or it is no caching)
2008-06-24T12:59:47 <johill> oh ok I thought it was serving data from another cache
2008-06-24T13:00:09 <ThomasWaldmann> e.g. a thumbnail macro would feed the data into that data cache
2008-06-24T13:00:21 <ThomasWaldmann> (and the header stuff into the meta cache)
2008-06-24T13:00:36 <johill> ok
2008-06-24T13:01:31 <johill> so I think this is useful only if we either add an ACL check (probably hard to do it would have to call back to the generator somehow?)
2008-06-24T13:01:44 <johill> or we use this for unprotected content only
2008-06-24T13:01:58 <johill> and add a separate thing that uses the session for the keys
2008-06-24T13:03:45 <johill> for protected content
2008-06-24T13:04:56 <ThomasWaldmann> well, strictly taken we have no unprotected content
2008-06-24T13:05:20 <johill> yeah, that's true
2008-06-24T13:05:33 <johill> but then the feature would require anon sessions
2008-06-24T13:06:21 <ThomasWaldmann> hmm, an acl check in the data sending branch is maybe not that critical as for most views it should be the 304 branch anyway
2008-06-24T13:06:53 <johill> the thing is, how do you do the ACL check?
2008-06-24T13:06:57 <ThomasWaldmann> (timing and load critical I mean)
2008-06-24T13:07:14 <johill> it would have to have a pagename of which you check the ACLs
2008-06-24T13:07:21 <ThomasWaldmann> one could write the complete acl into the meta dict, too :P
2008-06-24T13:07:46 <johill> no
2008-06-24T13:08:07 <johill> that has the revoke problem again
2008-06-24T13:08:21 <ThomasWaldmann> hmm, right
2008-06-24T13:08:46 <ThomasWaldmann> then it has to point to the acl source
2008-06-24T13:09:15 <johill> yeah
2008-06-24T13:09:25 <ThomasWaldmann> (currently the attachment's parent Page name, later the item name of the attachment)
2008-06-24T13:09:42 <johill> yeah something like that
2008-06-24T13:10:07 <johill> it'd be simpler if we generated one-time keys on the fly
2008-06-24T13:10:22 <johill> but I guess that defeats browser caching
2008-06-24T13:10:25 <ThomasWaldmann> gtg/bbl
2008-06-24T13:10:29 <johill> so not useful
2008-06-24T13:13:17 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3788:8ddd2b1c49b4 1.8/MoinMoin/macro/WikiConfigHelp.py: WikiConfigHelp: translate and mark up descriptions
2008-06-24T13:16:14 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3789:9173925a53da 1.8/MoinMoin/config/multiconfig.py: config: move page_license_page
2008-06-24T13:23:13 <gizmach> re
2008-06-24T13:40:10 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3790:4340777f31f9 1.8/ (3 files in 3 dirs):
2008-06-24T13:40:10 <CIA-52> remove backup action
2008-06-24T13:40:10 <CIA-52> This code is quite unsafe and is totally incompatible with the
2008-06-24T13:40:10 <CIA-52> storage abstraction.
2008-06-24T13:48:49 <ThomasWaldmann> johill: about the 'revoking rights' problem again
2008-06-24T13:49:32 <ThomasWaldmann> if someone got the url of object x and the url is dependant on content of x, then he already knows (or can know) the content and could have copied it
2008-06-24T13:49:51 <ThomasWaldmann> so strictly taken we don't need to revoke access to the same content
2008-06-24T13:50:09 <ThomasWaldmann> and if key depends on content, url will change if content changes
2008-06-24T14:01:27 <johill> true
2008-06-24T14:01:47 <johill> still seems a bit odd
2008-06-24T14:03:46 <xorAxAx> hmm, no student today
2008-06-24T14:04:46 * johill leaves too
2008-06-24T14:11:33 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4042:698b8e0204d3 1.8-mercurialbackend-ppacana/MoinMoin/storage/ (__init__.py backends/hg.py backends/memory.py abstract.py): Merge with 1.8-storage-cdenter
2008-06-24T14:11:33 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4047:4c278f90b459 1.8-mercurialbackend-ppacana/MoinMoin/storage/ (backends/memory.py error.py _tests/test_backends_memory.py): Merge with 1.8-storage-cdenter
2008-06-24T14:11:41 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4061:9c6d27fb53e9 1.8-mercurialbackend-ppacana/MoinMoin/storage/ (3 files in 3 dirs): Merge with 1.8-storage-cdenter
2008-06-24T14:11:41 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4062:814e42f254f7 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/ (test_backends.py test_backends_hg.py): Adopted Christopher memory backend tests - made it more generic for subclassing in other backend tests. Subclassed BackendTest for MercurialBackend.
2008-06-24T14:11:43 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4063:cf61caf43748 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py: Added has_item, _create_revision. Improper locking fixes. Bunch of refactorings.
2008-06-24T14:11:46 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4066:cef76d0b0439 1.8-mercurialbackend-ppacana/MoinMoin/storage/ (3 files in 3 dirs): Merge with 1.8-storage-cdenter
2008-06-24T14:11:48 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4067:3035b88ec60b 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends.py: BackendTest tests follow-up from MemoryBackend tests.
2008-06-24T14:13:08 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3791:7f43f5c38310 1.8/MoinMoin/macro/WikiConfigHelp.py: WikiConfigHelp macro: hide default already at 30 characters, use span to show on mouse hover
2008-06-24T14:26:38 <gizmach> xorAxAx: ?
2008-06-24T14:28:51 <xorAxAx> gizmach: i mean byeongweon :)
2008-06-24T14:29:04 <xorAxAx> who was here only for 36 seconds today :)
2008-06-24T14:29:06 <gizmach> ah :)
2008-06-24T14:29:18 <gizmach> well better than nothing
2008-06-24T14:31:10 <xorAxAx> gizmach: hehe
2008-06-24T14:31:20 <xorAxAx> how is the weather in croatia?
2008-06-24T14:38:24 <gizmach> sunny and hot
2008-06-24T14:38:37 <gizmach> in germany?
2008-06-24T14:48:31 <xorAxAx> gizmach: 21°
2008-06-24T14:49:00 <gizmach> uh in my appartment is 28
2008-06-24T14:49:07 <gizmach> 28,5 pardon
2008-06-24T14:49:17 <TheSheep> gizmach: celsius?
2008-06-24T14:49:31 <gizmach> yes
2008-06-24T14:49:41 <TheSheep> neat, my favorite temperature
2008-06-24T14:50:33 <xorAxAx> gizmach: here its 32° in the appartment :)
2008-06-24T14:50:50 <gizmach> ah
2008-06-24T14:50:55 <gizmach> TheSheep: mine not
2008-06-24T14:51:05 <gizmach> It's to hot to think
2008-06-24T14:51:32 <TheSheep> if ingorance is bliss, 'tis fooly to be wise :)
2008-06-24T14:52:23 <TheSheep> just enjoy the sun
2008-06-24T14:58:46 <gizmach> TheSheep: hehe
2008-06-24T15:15:10 <dennda> johill: seen my questions? :)
2008-06-24T15:37:46 <zenhase> i guess my room has about 35, luckily i am not there
2008-06-24T15:52:45 <CIA-52> Radomir Dopieralski <moindev@sheep.art.pl> default * 4068:cf9f8fc47693 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends.py: formatting and variable renames in test_backends.py
2008-06-24T16:21:16 <johill> dennda: sort of ;)
2008-06-24T16:24:42 <TheSheep> what happens when two instances of moin decide to call create_item at the same moment?
2008-06-24T16:24:47 <johill> dennda: PawelPacana: ok one of the questions was whether create_item fails quickly. ISTR thinking create_item shouldn't create any state at all
2008-06-24T16:25:08 <johill> until metadata/revisions are added
2008-06-24T16:27:36 <TheSheep> and commited/unlocked
2008-06-24T16:28:17 <TheSheep> johill: is the metadata lock a rw lock or just write?
2008-06-24T16:28:28 <johill> just write
2008-06-24T16:28:36 <johill> it's badly named as a lock though
2008-06-24T16:28:49 <johill> I'm still trying to think of a better idiom
2008-06-24T16:28:59 <johill> we want to lock + save atomically on unlock
2008-06-24T16:29:03 <TheSheep> commit :)
2008-06-24T16:29:10 <johill> yeah but lock + commit?
2008-06-24T16:29:29 <TheSheep> npo, commit metadata and commit data
2008-06-24T16:29:41 <TheSheep> name it "howgh"
2008-06-24T16:29:53 <TheSheep> or so_be_it
2008-06-24T16:30:00 <TheSheep> or "amen"
2008-06-24T16:30:24 <johill> well it can be named anything, but what captures the semantics best?
2008-06-24T16:30:34 <TheSheep> johill: amen :)
2008-06-24T16:30:40 <johill> :)
2008-06-24T16:31:08 <johill> dennda: anyway, yes, we need to lock manually
2008-06-24T16:33:41 <johill> the only question is how to capture the semantics of lock/commit properly
2008-06-24T16:34:14 <johill> maybe .edit/.publish
2008-06-24T16:34:28 <johill> (not sure if that's a serious suggestion or not)
2008-06-24T16:34:42 <TheSheep> sounds good
2008-06-24T16:34:53 <TheSheep> not sure about .edit
2008-06-24T16:35:05 <TheSheep> doesn't soundlocky ;)
2008-06-24T16:35:10 <TheSheep> sound locky
2008-06-24T16:35:14 <johill> true
2008-06-24T16:35:24 <TheSheep> maybe open?
2008-06-24T16:35:26 <TheSheep> and close
2008-06-24T16:35:36 <TheSheep> yeah, bad idea
2008-06-24T16:37:57 <johill> like I said, this is the problem :)
2008-06-24T16:40:13 <TheSheep> how about the non sequitur gambit?
2008-06-24T16:41:01 <TheSheep> the question is who is to be the master
2008-06-24T16:41:10 <johill> hm?
2008-06-24T16:42:41 <TheSheep> johill: http://sheep.art.pl/HumptyDumpty
2008-06-24T16:44:29 <johill> google-analytics.com is making it drag here
2008-06-24T16:44:38 <TheSheep> ugh
2008-06-24T16:44:43 <TheSheep> I need to remove it anyways
2008-06-24T16:44:53 <TheSheep> thanks for reminding me :)
2008-06-24T16:46:29 <johill> ah, loaded now :)
2008-06-24T16:47:24 <johill> well, yeah, of course :)
2008-06-24T17:05:41 <johill> TheSheep, PawelPacana: I take it you are merging dennda's branch? should we merge 1.8 in dennda's branch or keep doing things in hwendel?
2008-06-24T17:05:55 <johill> (thing is that one is called 1._7_-storage-hwendel)
2008-06-24T17:06:36 <TheSheep> johill: I think that ThomasWaldmann merged hwendel, at least a large part of it, into 1.8?
2008-06-24T17:07:02 <johill> no
2008-06-24T17:07:21 <johill> nothing at all, and I don't think we should
2008-06-24T17:08:57 <TheSheep> PawelPacana: ping
2008-06-24T17:10:08 <johill> since we're not going to merge anything storage-related into 1.7, I'd just continue using 1.7-storage-hwendel
2008-06-24T17:10:12 <johill> but the naming would be wrong
2008-06-24T17:10:25 <johill> (so much for naming topic branches after versions, why anyway?)
2008-06-24T17:12:16 <TheSheep> so that you don't have this "moin-new-extra-cool-feature" with last commit from 1972?
2008-06-24T17:12:46 <johill> heh
2008-06-24T17:13:04 <johill> dreimark: http://hg.moinmo.in/moin/1.8/rev/749d7148b080 -> you know there is a setting to disable the duplicate email check? not sure, I think we should possibly removme it
2008-06-24T17:13:08 <PawelPacana> johill: we`re merging dennda`s
2008-06-24T17:42:51 <johill> ok I'll think about it
2008-06-24T17:53:42 <dreimark> johill: in create_account if user.get_by_email_address(request, self.options.email): ?
2008-06-24T17:55:49 <johill> if theuser.email and request.cfg.user_email_unique:
2008-06-24T17:56:09 <dreimark> ThomasWaldmann: is the key changed if only the attachment is overwritten and how does it change the key if one removes one from a group. Then there is no content change.
2008-06-24T17:58:11 <gizmach> dreimark: hi :)
2008-06-24T18:00:00 <dreimark> hi melita
2008-06-24T18:03:03 <dreimark> johill: yeah it should be same as in newaccount
2008-06-24T18:09:51 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4069:55a95c367d61 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends_hg.py: Added fresh repository creation for every hg backend test.
2008-06-24T18:09:51 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4070:dcd7c949097f 1.8-mercurialbackend-ppacana/MoinMoin/storage/_tests/test_backends.py: Naming convention misconception.
2008-06-24T18:09:52 <CIA-52> Pawel Pacana <pawel.pacana@gmail.com> default * 4071:e3917b38be98 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py: Docstring and formatting fixes
2008-06-24T18:23:48 <CIA-52> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3792:4f1d2cd9879b 1.8/MoinMoin/script/account/create.py: script.account.create: use same rule for email check than in MoinMoin.action.newaccount
2008-06-24T18:24:54 <johill> pootle looks nice
2008-06-24T18:25:44 <dreimark> http://translate.sourceforge.net/wiki/pootle/index
2008-06-24T18:26:16 <johill> yeah
2008-06-24T18:30:55 <ThomasWaldmann> dreimark: the point of my reasoning above is that you don't have to hide content the user could get before removing him by acl
2008-06-24T18:31:21 <ThomasWaldmann> (because he can already know that content, could've copied it etc.)
2008-06-24T18:31:59 <ThomasWaldmann> the key would have to be something like sha(secret+content) in that case
2008-06-24T18:32:52 <ThomasWaldmann> just content would maybe also work for that case
2008-06-24T18:33:05 <ThomasWaldmann> sha(content) I mean
2008-06-24T18:33:14 <dreimark> the key is not based on the attachment. if the page changes the key is changed.
2008-06-24T18:33:37 <dreimark> but you won't do that for the attachment that is to expensive
2008-06-24T18:33:38 <ThomasWaldmann> no, the key is based on the attachment
2008-06-24T18:34:16 <ThomasWaldmann> ehrm, no
2008-06-24T18:35:05 <ThomasWaldmann> the key is based on the source material for the stuff we rendered and stored to the cache
2008-06-24T18:35:28 <ThomasWaldmann> no matter whether that source is an attachment, or some lines of text in a parser section or whatever
2008-06-24T18:36:22 <ThomasWaldmann> for an image thumbnail, it would be the image (for large images sha could take a bit to compute, no idea how long?)
2008-06-24T18:37:08 <ThomasWaldmann> for a latex formula, it would be the latex src (and the cache content would be a png rendered from it)
2008-06-24T18:37:49 <johill> remember to also have an easy way to delete those items again
2008-06-24T18:38:01 <ThomasWaldmann> rm -rf :)
2008-06-24T18:38:05 <johill> if somebody changes the latex formula you don't necessarily know what the original was
2008-06-24T18:38:07 <dreimark> that can become expensive and it depends on the size of the data
2008-06-24T18:38:22 <dreimark> while the usual acl check does need always quite the same time
2008-06-24T18:39:27 <johill> rm is necessary even with acl checks
2008-06-24T18:39:39 <ThomasWaldmann> yes, it depends on data size (assuming we read whole file what we maybe don't need)
2008-06-24T18:40:49 <ThomasWaldmann> we could also compute the has from other stuff, esp for files
2008-06-24T18:40:54 <ThomasWaldmann> +h
2008-06-24T18:41:21 <ThomasWaldmann> (e.g. item name, file length, mtime)
2008-06-24T18:42:19 * ThomasWaldmann .oO(fuid :)
2008-06-24T18:42:24 <dreimark> we learned already lot's of mtime
2008-06-24T18:43:20 <ThomasWaldmann> if you manage to update the file 2times per second, keeping the name and the length, you would get the same key
2008-06-24T18:43:29 <dreimark> well currently I would really prefer acls and on a further Attachfile upload on upload creating of an md5sum
2008-06-24T18:43:58 <dreimark> We do use this since some time for filesystem copies
2008-06-24T18:44:04 <ThomasWaldmann> attachfile for storing cache content is crap
2008-06-24T18:44:05 <dreimark> and there are regular files broken
2008-06-24T18:44:34 <dreimark> md5sum has to go into the metadata
2008-06-24T18:44:48 <ThomasWaldmann> (and will be gone soon anyway)
2008-06-24T18:45:59 <xorAxAx> ThomasWaldmann: you seem to have talked about a MAC
2008-06-24T18:46:11 * ThomasWaldmann also thought of storing a md5sum/sha for items, but first we need items :)
2008-06-24T18:46:48 <ThomasWaldmann> I'll play with a MacBook tomorrow :)
2008-06-24T18:47:11 <zenhase> :)
2008-06-24T18:47:18 <zenhase> what do you intend to do with it?
2008-06-24T18:47:32 <ThomasWaldmann> sell it :)
2008-06-24T18:48:01 <ThomasWaldmann> (after putting 4Gigs in it :)
2008-06-24T18:48:46 <dreimark> johill: btw. some configuration options for a wiki should be from within a SystemAdmin macro
2008-06-24T18:49:08 <dreimark> e.g. disabling specific newaccount creation
2008-06-24T18:49:17 <dreimark> not direct actions_ecluded
2008-06-24T18:49:29 <dreimark> only some few options
2008-06-24T18:50:06 <johill> hm?
2008-06-24T18:50:11 <johill> you mean should be settable
2008-06-24T18:50:12 <johill> ?
2008-06-24T18:50:19 <dreimark> yeah
2008-06-24T18:50:37 <johill> that's pretty much impossible
2008-06-24T18:50:44 <johill> you'd have to rewrite the config file
2008-06-24T18:50:52 <johill> or have a second place where config is stored which is also very undesirable
2008-06-24T18:50:58 <dreimark> ok
2008-06-24T18:51:16 <dreimark> but it is boring to add a new user to a wiki where the action is disabled
2008-06-24T18:51:25 <dreimark> you have to kill the server process ...
2008-06-24T18:51:48 <johill> oh
2008-06-24T18:51:56 <johill> well I suppose it would be possible to set it temporarily
2008-06-24T18:52:34 <johill> not for cgi, of course
2008-06-24T18:52:44 <johill> and that means it's trouble to support different installations
2008-06-24T18:52:57 <johill> so I suggest not to do it ;)
2008-06-24T18:53:41 <johill> dreimark: what you could do is extend textcha support to have per-action textchas, ie. newaccount can have different ones than edit; and then you can disable them for newaccount and use them to have a custom password
2008-06-24T18:53:50 <johill> or something like that
2008-06-24T18:53:55 <dreimark> or may be a xmlrpc call for user creation
2008-06-24T18:56:08 <dreimark> or a form for usercreations from superuser
2008-06-24T18:56:18 <johill> all well doable
2008-06-24T18:56:18 <dreimark> so that ther is one user who can create users
2008-06-24T18:59:02 <CIA-52> Johannes Berg <johannes AT sipsolutions DOT net> default * 3795:6e73d9be3e3f 1.8/MoinMoin/ (3 files in 3 dirs): merge 1.7
2008-06-24T19:21:16 <dreimark> bbl
2008-06-24T19:40:15 <xorAxAx> ThomasWaldmann: mac != MAC
2008-06-24T19:42:00 <johill> xorAxAx: for that matter, MAC != MAC ;)
2008-06-24T19:42:33 <xorAxAx> johill: why?
2008-06-24T19:42:40 <xorAxAx> MAC is a general term
2008-06-24T19:42:54 <ThomasWaldmann> xorAxAx: even if you mean the other MAC, i can't make much from your comment :)
2008-06-24T19:43:03 <johill> for a million things
2008-06-24T19:43:07 <xorAxAx> ThomasWaldmann: you talked about hashing of data
2008-06-24T19:43:10 <ThomasWaldmann> I did. Yes, AND?
2008-06-24T19:43:14 <xorAxAx> describe how its different from a mac :)
2008-06-24T19:43:36 <johill> huh?
2008-06-24T19:43:40 <johill> now I'm confused
2008-06-24T19:43:45 <johill> what is a mac?
2008-06-24T19:43:51 <TheSheep> describe the difference between DHCP and AGP
2008-06-24T19:43:52 <ThomasWaldmann> johill: MAC like in HMAC
2008-06-24T19:44:17 <ThomasWaldmann> (not iMAC :P )
2008-06-24T19:44:20 <johill> well in upper case MAC is message authentication code (or media access controller or mandatory access control or...)
2008-06-24T19:44:25 <TheSheep> ^-- that's one of the questions we ask candidates for new admins
2008-06-24T19:44:58 <xorAxAx> TheSheep: and what do you expect? :)
2008-06-24T19:45:04 <johill> TheSheep: umm?
2008-06-24T19:45:18 <johill> xorAxAx: so what is a lower-case mac? a word for a macintosh computer?
2008-06-24T19:45:26 <ThomasWaldmann> TheSheep: like "Tell us 5 things what a MAC is"?
2008-06-24T19:45:50 <TheSheep> ThomasWaldmann: no, that DHCP vs. AGP thing :)
2008-06-24T19:45:54 <xorAxAx> johill: no, a lazily written version of the upper case one
2008-06-24T19:46:03 <xorAxAx> ThomasWaldmann: so why do you need such a hash?
2008-06-24T19:46:04 <TheSheep> xorAxAx: I'd expect MU
2008-06-24T19:46:21 <TheSheep> xorAxAx: but many people are unable to not answer
2008-06-24T19:46:24 <xorAxAx> TheSheep: MU?
2008-06-24T19:46:27 <xorAxAx> ah, the japanese one
2008-06-24T19:46:34 <TheSheep> chinese actually
2008-06-24T19:46:37 <TheSheep> but yes
2008-06-24T19:46:58 <ThomasWaldmann> we need a key into the cache that is transmitted in the action url
2008-06-24T19:47:07 <zenhase> eeeek ... raindrops on my display :o
2008-06-24T19:47:11 <zenhase> gotta change location
2008-06-24T19:47:51 <ThomasWaldmann> and as the cache content might be not for everyone's eyes, the key should not be easily predictable
2008-06-24T19:48:34 <ThomasWaldmann> and change if the render source changes
2008-06-24T19:52:31 <ThomasWaldmann> gtg/bbl
2008-06-24T20:25:37 <dennda> johill: Ok, so a) I need to make create_item just return an Item and commit() handle the potential clashing and b) rename _lock and _unlock (why the underscores anyway?)
2008-06-24T20:26:37 <johill> I think the underscores were based on a misunderstanding
2008-06-24T20:26:55 <johill> but I think so yes
2008-06-24T20:27:11 <dennda> Did you review my latest changesets?
2008-06-24T20:27:37 <johill> good question
2008-06-24T20:28:48 <johill> looks fine
2008-06-24T20:29:06 <johill> I think you forgot metadata updates though
2008-06-24T20:29:06 <dennda> ok
2008-06-24T20:29:16 <dennda> lemme check
2008-06-24T20:29:30 <johill> it looks as though changes never get written in the mem backend
2008-06-24T20:31:05 <johill> you'll see when you start writing metadata tests :)
2008-06-24T20:33:23 <dennda> metadata updates? you mean first creating one metadata key-value-pair and then altering that keys value?
2008-06-24T20:34:25 <johill> saving metadata at all
2008-06-24T20:34:50 <johill> oh I see it's actually wrong
2008-06-24T20:35:20 <xorAxAx> ThomasWaldmann: ah
2008-06-24T20:37:46 <dennda> http://paste.pocoo.org/show/77623/
2008-06-24T20:38:03 <dennda> johill: What's wrong with that?
2008-06-24T20:38:17 <johill> that's why I was saying lock/unlock is misnamed
2008-06-24T20:38:27 <johill> metadata changes may not be visible to concurrent readers until unlock
2008-06-24T20:39:07 <johill> otherwise they can get inconsistent data
2008-06-24T20:39:27 <dennda> i see
2008-06-24T20:39:38 <johill> any better names?
2008-06-24T20:39:56 <johill> change_metadata ... publish_metadata?
2008-06-24T20:40:08 <dennda> so you want them to either get the value that was there before the lock (trivial) or after unlock
2008-06-24T20:40:51 <johill> yeah so that it won't have half-updated data
2008-06-24T20:44:10 <dennda> change_metadata and publish_metadata is good imho
2008-06-24T20:48:50 <dennda> johill: I think how to best implement it, what do you think: upon altering metadata, store a copy of the old metadata dict. if one asks for a key in the meantime, return it from the old dict. after publishing the old metadata dict object is set to none and __getitem__ uses the normal metadata dict
2008-06-24T20:49:35 <ThomasWaldmann> johill: I suggest you remove backup action in storage branch (or after merging storage branch), not now in main branch
2008-06-24T20:50:01 <johill> ThomasWaldmann: why not? it's not really working anyway
2008-06-24T20:50:15 <johill> dennda: you only need to change a single line
2008-06-24T20:50:15 <ThomasWaldmann> what is not working?
2008-06-24T20:50:20 <johill> backup
2008-06-24T20:50:28 <johill> dennda: return self._item_metadata[item._item_id]
2008-06-24T20:50:32 <ThomasWaldmann> be more precise :)
2008-06-24T20:50:36 <johill> dennda: -> return dict(self._item_metadata[item._item_id])
2008-06-24T20:50:52 <johill> ThomasWaldmann: there's a huge warning about it, that's all really
2008-06-24T20:51:20 <johill> ThomasWaldmann: I don't mind keeping it, but that warning has been there for a long time so evidently nobody cares
2008-06-24T20:51:22 <ThomasWaldmann> I wrote that warning because it is not heavily tested
2008-06-24T20:52:02 <ThomasWaldmann> I see no reason for removing it early.
2008-06-24T20:52:22 <dennda> phone
2008-06-24T20:52:25 <ThomasWaldmann> If we have to, we will, but we still have old storage code in main branch.
2008-06-24T20:53:22 <johill> ThomasWaldmann: different operating principle I guess, I see that nobody uses it and nobody cares, and if anybody wants the code is still there, but right now it's more of a maintenance burden imo
2008-06-24T20:53:48 <ThomasWaldmann> how do you know nobody uses it? i don't know that.
2008-06-24T20:54:18 <johill> I'd think we'd have heard about that
2008-06-24T20:55:00 <ThomasWaldmann> if it were non-working, yes, if it works, no.
2008-06-24T20:55:29 <ThomasWaldmann> so, if you tested it and you found it broken, you are right.
2008-06-24T20:55:48 <ThomasWaldmann> if you didn't test it, but just removed it early, you are wrong.
2008-06-24T20:55:59 <johill> I only got as far as not seeing the UI because I didn't find the settings ;)
2008-06-24T20:56:12 <johill> however, clearly nobody cares to maintain it
2008-06-24T20:57:02 <ThomasWaldmann> well, I wrote it for people without ssh or other direct server access
2008-06-24T20:57:25 <ThomasWaldmann> I don't use it myself.
2008-06-24T20:57:25 <johill> by all means, if you're attached to it, put it back
2008-06-24T20:57:45 <johill> but having code in there that isn't maintained is just wrong in my opinion
2008-06-24T20:58:17 <ThomasWaldmann> we can try waiting until someone complains :)
2008-06-24T20:58:38 <ThomasWaldmann> we have lots of badly maintained code
2008-06-24T20:58:59 <ThomasWaldmann> usually the stuff that the devs don't use themselves
2008-06-24T20:59:07 <johill> heh that's a plan too
2008-06-24T20:59:12 <johill> yeah, that's a problem
2008-06-24T20:59:56 <ThomasWaldmann> otoh, that doesn't mean other people don't use it (see docbook, see RSS, see RST, ...)
2008-06-24T21:00:57 <ThomasWaldmann> btw, do you intend to flag all descriptions with a dummy _()?
2008-06-24T21:01:07 <johill> I haven't decided yet
2008-06-24T21:01:13 <johill> I don't want to collect them into the same i18n file
2008-06-24T21:01:16 <johill> po file I mean
2008-06-24T21:01:25 <dreimark> at least there should be unit tests for it that we know it got's (more) broken by some changes
2008-06-24T21:01:26 <johill> because those descriptions are clearly much lower priority to translate
2008-06-24T21:01:42 <ThomasWaldmann> would be a lot of extra work for translators
2008-06-24T21:02:06 <ThomasWaldmann> (and maybe not every translator sees a need to translate that, like me :)
2008-06-24T21:02:23 <johill> exactly
2008-06-24T21:02:32 <johill> so we should put them into a separate po file
2008-06-24T21:02:48 <johill> and between dreimark, xorAxAx and me we never figured out how exactly texts are collected ;)
2008-06-24T21:02:49 <ThomasWaldmann> no infrastructure yet for that
2008-06-24T21:03:12 <ThomasWaldmann> using xgettext
2008-06-24T21:03:44 <johill> ok, I was confused by check_i18n for a while
2008-06-24T21:05:54 <ThomasWaldmann> it's all in the Makefile
2008-06-24T21:06:30 <dreimark> may be can use some of this code http://wiki.wxpython.org/Internationalization#head-a5665108d7852c4069838327388fc11fbc524dfb
2008-06-24T21:07:29 <dreimark> ThomasWaldmann: it takes time to found the Makefile and the README because I was searching in .py
2008-06-24T21:07:45 <dreimark> and was not aware that there is a README in that i18n dir
2008-06-24T21:08:43 <ThomasWaldmann> that source looks like they didn't even get the line endings right
2008-06-24T21:09:13 <dreimark> and lots of pep8
2008-06-24T21:10:00 <dreimark> we should have a syntax highlighter which uses python.vim
2008-06-24T21:11:50 <dreimark> TheSheep: what problem has FF2 ?
2008-06-24T21:12:21 <ThomasWaldmann> strange modernized rendering
2008-06-24T21:13:34 <ThomasWaldmann> johill: about that transient fields question maybe look when the cfg setting was introduced
2008-06-24T21:13:57 <ThomasWaldmann> I can't remember why, but IIRC it was just a attribute in user.py before.
2008-06-24T21:15:14 <ThomasWaldmann> and don't touch the user_checkbox* stuff except for fixing bugs
2008-06-24T21:15:31 <ThomasWaldmann> some admins need that stuff
2008-06-24T21:15:49 <ThomasWaldmann> (same for user_form*)
2008-06-24T21:19:16 <johill> yeah but a bunch of the user_form* things are probably not used any more
2008-06-24T21:27:19 <TheSheep> dreimark: min-width in a floated element is 0
2008-06-24T21:46:00 * ThomasWaldmann .oO(good we have no default user with default password, see heise ticker :)
2008-06-24T21:47:36 <TheSheep> which unineligible German title exactly? :)
2008-06-24T21:50:06 <TheSheep> " A final evaluation would not Schurig, as the investigation is still ongoing." <-- I love google translate
2008-06-24T21:57:17 <ThomasWaldmann> TheSheep: some german cities had a big data leak with personal data of citizens because someone forgot to change the default pw of the web app
2008-06-24T22:00:30 <TheSheep> you are only as safe as your worst admin
2008-06-24T22:01:51 <TheSheep> ok, I'm copying the modernized theme into moin-1.8 repo
2008-06-24T22:04:39 <ThomasWaldmann> did you find the ff2 problem?
2008-06-24T22:04:49 <TheSheep> yes
2008-06-24T22:05:04 <ThomasWaldmann> looks better now
2008-06-24T22:05:09 <TheSheep> ff2 sets max-width on floated elements to 0
2008-06-24T22:05:40 <ThomasWaldmann> btw, if that rss icon could be somewhere else, that would be welcome
2008-06-24T22:05:53 <TheSheep> f actually has a lot of rendering bugs too, especially with inserted elements
2008-06-24T22:06:11 <TheSheep> ThomasWaldmann: for example where?
2008-06-24T22:06:23 <ThomasWaldmann> where it is now, it takes a full line of screenspace
2008-06-24T22:07:15 <ThomasWaldmann> maybe below the RC data, right of the icon table?
2008-06-24T22:08:24 <TheSheep> I can try to put it on top of the table
2008-06-24T22:09:08 <ThomasWaldmann> if it is above all the other icons, that is maybe a bit stealthy
2008-06-24T22:10:23 <TheSheep> hmm.. interesting
2008-06-24T22:10:27 <TheSheep> look now :)
2008-06-24T22:11:03 <ThomasWaldmann> no please :)
2008-06-24T22:12:05 <TheSheep> I didn't know one could do that actually
2008-06-24T22:14:22 <TheSheep> how about now? :)
2008-06-24T22:15:47 <TheSheep> any other place would require me to change the actual html...
2008-06-24T22:23:24 <ThomasWaldmann> good :)
2008-06-24T22:24:56 <ThomasWaldmann> doesn't work perfect for rtl, though
2008-06-24T22:25:12 <TheSheep> ah, right
2008-06-24T22:25:26 <TheSheep> and doesn't work at all on msie6 -- it gets an empty line, but I guess we can live with that
2008-06-24T22:26:06 <ThomasWaldmann> but header, footer and RC table still look ok for RTL (afaics)
2008-06-24T22:26:30 <TheSheep> yeah, I fixed them yesterday ;)
2008-06-24T22:27:15 <ThomasWaldmann> (the real test would be some native rtler testing it)
2008-06-24T22:27:30 <dennda> johill: How's that supposed to fix the problem?
2008-06-24T22:28:08 <dennda> That won't even work but raise a ValueError
2008-06-24T22:29:23 <TheSheep> ThomasWaldmann: surprisingly, I don't know any
2008-06-24T22:29:32 <TheSheep> ok, should be fixed for rtl
2008-06-24T22:29:50 <johill> dennda: hm?
2008-06-24T22:30:08 <johill> oh the dict?
2008-06-24T22:30:18 <johill> you can pass a dict to the dict constructor, no?
2008-06-24T22:30:19 <ThomasWaldmann> TheSheep: cool, you're so fast with CSS stuff :)
2008-06-24T22:40:49 <CIA-52> Radomir Dopieralski <moindev@sheep.art.pl> default * 3796:668a7f2f1775 1.8/ (71 files in 3 dirs):
2008-06-24T22:40:49 <CIA-52> Introducing a remake of the default theme: Modernized
2008-06-24T22:40:49 <CIA-52> * a new set of tango icons
2008-06-24T22:40:49 <CIA-52> * improved colors of the header
2008-06-24T22:40:49 <CIA-52> * better display with CSS disabled
2008-06-24T22:40:50 <CIA-52> * minor spacing and border adjustments
2008-06-24T22:40:52 <CIA-52> * cleaner css for the header
2008-06-24T22:41:56 <ThomasWaldmann> hey, moin has a good looking theme \o/
2008-06-24T22:42:08 <ThomasWaldmann> (builtin theme I mean :)
2008-06-24T22:42:14 <johill> what about tango icon licensing though? do we have to be super-careful?
2008-06-24T22:43:14 <ThomasWaldmann> if there is some license text missing, we have docs/licenses/
2008-06-24T22:51:29 <dennda> johill: except that that isn't a dict but a value of the dict
2008-06-24T22:51:45 <dennda> but maybe I am missing the point, my brain is a bit idle at the moment
2008-06-24T22:52:35 <johill> dennda: for one, you defined _get_item_metadata twice
2008-06-24T22:52:49 <johill> def _get_item_metadata(self, item):
2008-06-24T22:52:49 <johill> """
2008-06-24T22:52:49 <johill> Load metadata for a given item, return dict.
2008-06-24T22:52:49 <johill> """
2008-06-24T22:52:49 <johill> return self._item_metadata[item._item_id]
2008-06-24T22:53:04 <johill> so if that returns a dict it can well return dict(...)
2008-06-24T22:53:43 <TheSheep> johill: these guys are usually not into suing everything around, but good point
2008-06-24T22:55:30 <johill> TheSheep: true, but still, debian for example doesn't ship any such icons in free IIRC
2008-06-24T22:59:56 <TheSheep> johill: because they are not GPL
2008-06-24T23:00:15 <johill> yeah
2008-06-24T23:00:24 <TheSheep> hmm.. would that mean they would not ship moinmoin then?
2008-06-24T23:00:24 <johill> well, more precisely, because they aren't dfsg-free
2008-06-24T23:00:35 <johill> sometimes they strip stuff out
2008-06-24T23:01:46 <CIA-52> Radomir Dopieralski <moindev@sheep.art.pl> default * 3797:fade016565c0 1.8/docs/licenses/modernized.icons.txt: License for the Tango icons in the Modernized theme
2008-06-24T23:04:13 <dreimark> good night
2008-06-24T23:15:07 <ThomasWaldmann> gn dreimark
2008-06-24T23:42:41 <ThomasWaldmann> TheSheep: the stemmer makes modern from modernized :P
2008-06-24T23:44:12 <ThomasWaldmann> http://moinmo.in/ThemeMarket/Modernized < maybe update the package there also
2008-06-24T23:44:52 <ThomasWaldmann> if this gets in broader use (including moinmo.in site), we can have it better tested than in 1.8dev code
2008-06-24T23:53:57 <ThomasWaldmann> no, 10:15
2008-06-24T23:54:04 <ThomasWaldmann> ignore :) MoinMoin: MoinMoinChat/Logs/moin-dev/2008-06-24 (last edited 2008-06-23 22:30:02 by IrcLogImporter)
