MoinMoin: MoinMoinChat/Logs/moin-dev/2008-06-08

2008-06-08T00:00:09  <CIA-50> MelitaMihaljevic default * 3674:1815a4df41c6 1.8-ldapgroups-mmihaljevic/MoinMoin/_tests/test_ldapgroups.py: new tests for ldap
2008-06-08T00:09:06  <gizmach> good night
2008-06-08T00:14:44  <ThomasWaldmann> gizmach: is your keyboard defective somehow?
2008-06-08T00:16:50  <ThomasWaldmann> gizmach: tomorrow please review your code (it is overly complicated), add comments, make it pep8 compliant
2008-06-08T01:30:00  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3669:0cbd8b41aebd 1.7/ (MoinMoin/search/queryparser.py docs/CHANGES): Xapian search: fix searching for negative terms only
2008-06-08T04:12:04  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3670:78850075daad 1.7/MoinMoin/action/info.py: info action: fix exception happening for invalid query args
2008-06-08T06:16:55  <starshine_away> .33
2008-06-08T11:22:36  <ThomasWaldmann> moin
2008-06-08T11:22:43  <TheSheep> moin
2008-06-08T12:08:06  <zenhase> moin
2008-06-08T12:15:42  <CIA-50> Bastian Blank <bblank@thinkmo.de> default * 3646:afe613a7a40b 1.8-dom-bblank/MoinMoin/converter2/html_out.py:
2008-06-08T12:15:42  <CIA-50> Base converter.
2008-06-08T12:15:42  <CIA-50> MoinMoin/converter2/html_out.py
2008-06-08T12:15:42  <CIA-50> - Use specified function in recursor.
2008-06-08T12:15:42  <CIA-50> - Use dicts for namespace and tagname lookup.
2008-06-08T12:19:08  <zenhase> hmm ... that's what a multiline commit message looks like in CIA
2008-06-08T12:19:41  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3655:5bd2da7053a3 1.8-wsgi-fkrupicka/MoinMoin/ (3 files in 3 dirs): Removed httpDate from request object
2008-06-08T12:19:42  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3656:3b36f0cfc656 1.8-wsgi-fkrupicka/MoinMoin/ (11 files in 8 dirs): (Re)moved normalizePagename from Request into wikiutil library
2008-06-08T12:19:43  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3657:95e9e24da409 1.8-wsgi-fkrupicka/MoinMoin/web/contexts.py: Extended attribute/property access logging
2008-06-08T12:19:44  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3658:78e6deeadc7e 1.8-wsgi-fkrupicka/MoinMoin/web/ (contexts.py apps.py): Make XML-RPC action call into own WSGI app (draft)
2008-06-08T12:19:45  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3659:769850545997 1.8-wsgi-fkrupicka/MoinMoin/ (Page.py request/__init__.py wsgiapp.py): Removed request.editlog
2008-06-08T12:19:48  <CIA-50> Florian Krupicka <florian.krupicka@googlemail.com> default * 3660:72604cb46155 1.8-wsgi-fkrupicka/MoinMoin/web/utils.py: Just use the browser property of useragent object
2008-06-08T12:19:52  <zenhase> *cough*
2008-06-08T12:20:28  <dreimark> moin
2008-06-08T12:20:46  <zenhase> moin dreimark
2008-06-08T12:20:59  <dreimark> ThomasWaldmann: the devlog page is not active on MM
2008-06-08T12:21:05  <dreimark> hi zenhase
2008-06-08T12:22:14  <zenhase> ThomasWaldmann: was there any change to the mercurialguide? or did we use mercurial in a kind of wrong way?
2008-06-08T12:22:35  <zenhase> (referring to you mentioning the link yesterday)
2008-06-08T12:23:01  <dreimark> zenhase: there are some hints given for example to avoid two heads
2008-06-08T12:23:02  <TheSheep> that should be easy to check
2008-06-08T12:23:49  <dreimark> zenhase: look at ypur recent Changes and click the update icon
2008-06-08T12:24:34  <dreimark> Hi TheSheep, do you have read my ideas?
2008-06-08T12:30:53  <TheSheep> dreimark: yes, it's all possible
2008-06-08T12:31:51  <TheSheep> dreimark: even with partial transparency :)
2008-06-08T12:32:06  <dreimark> do you think it is a good idea to use the space right of more action ?
2008-06-08T12:32:26  <TheSheep> dreimark: I'm not sure what you mean
2008-06-08T12:33:25  <TheSheep> dreimark: what's 'more action' and where is it?
2008-06-08T12:34:12  <dreimark> in the edit bar the options list where we do have delete page ...
2008-06-08T12:34:14  <TheSheep> dreimark: I can also make the controls appear on hover
2008-06-08T12:34:30  <TheSheep> dreimark: ah, you mean some specific theme?
2008-06-08T12:34:38  <dreimark> sounds nice
2008-06-08T12:35:02  <dreimark> for the controls
2008-06-08T12:35:50  <dreimark> no not some specific theme, I do think what we can do with the space right of this option list
2008-06-08T12:36:14  <dreimark> TheSheep:
2008-06-08T12:37:53  <TheSheep> dreimark: other themes don't have that space
2008-06-08T12:38:25  <dreimark> ok, then we leave that idea
2008-06-08T12:38:42  <dreimark> s/leave/remove/
2008-06-08T12:39:32  * TheSheep makes a mockup
2008-06-08T12:51:27  <CIA-50> Bastian Blank <bblank@thinkmo.de> default * 3647:da786d5bc459 1.8-dom-bblank/MoinMoin/converter2/_tests/test_html_out.py:
2008-06-08T12:51:27  <CIA-50> Add real testcases. Fix exception.
2008-06-08T12:51:27  <CIA-50> MoinMoin/converter2/_tests/test_html_out.py
2008-06-08T12:51:27  <CIA-50> - Add basic testcase for sucessfull conversion.
2008-06-08T12:51:27  <CIA-50> - Check correct exception.
2008-06-08T12:51:29  <CIA-50> Bastian Blank <bblank@thinkmo.de> default * 3648:04d6517f8fef 1.8-dom-bblank/MoinMoin/converter2/html_out.py:
2008-06-08T12:51:34  <CIA-50> Remove type special casing.
2008-06-08T12:51:36  <CIA-50> MoinMoin/converter2/html_out.py
2008-06-08T12:51:38  <CIA-50> - Remove block and inline special casing.
2008-06-08T12:51:40  <CIA-50> - Fix to survive minimal tests.
2008-06-08T13:04:59  <TheSheep> dreimark: http://sheep.art.pl/misc/gallery2.html
2008-06-08T13:10:55  * dreimark looks
2008-06-08T13:15:25  <dreimark> TheSheep:
2008-06-08T13:15:39  <dreimark> opacity makes this transparency effect ?
2008-06-08T13:15:44  <TheSheep> yes
2008-06-08T13:16:31  <dreimark> where are the arraows defined
2008-06-08T13:16:50  <ThomasWaldmann> all students should please keep in mind that their commit comments should be understandable without context (not even with knowing what the current repo is about)
2008-06-08T13:17:11  <TheSheep> the <a class="next"> and <a clas="prev">, with arrows as text
2008-06-08T13:17:39  <ThomasWaldmann> so a comment like "add testcases" is pretty useless if it doesn't tell for what.
2008-06-08T13:17:39  <dreimark> ahh, I toally missed the -> in the text
2008-06-08T13:18:12  <dreimark> TheSheep:
2008-06-08T13:18:34  <TheSheep> dreimark: I could also make the whole navigation bar appear like this, I think
2008-06-08T13:19:14  <TheSheep> hmm... but not on hover, at least not without js in msie
2008-06-08T13:19:34  <dreimark> TheSheep: if you like you can play with the arnica code
2008-06-08T13:20:55  <dreimark> the old konqueror does not hover too, it is a black box
2008-06-08T13:21:43  <dreimark> an other idea (sorry don't have much experience with css effects)
2008-06-08T13:22:22  <dreimark> can all this tools be activated by a click or move into one of the edges
2008-06-08T13:22:40  <TheSheep> move -- yes, for click you'd need js
2008-06-08T13:22:49  <dreimark> so that the image can be viewed without conttrols
2008-06-08T13:23:15  <dreimark> (i do like to use it later for some of our model data output)
2008-06-08T13:23:16  <TheSheep> but I'd need js for msie anyways for this, as it can only do hover effects on links
2008-06-08T13:23:42  <dreimark> the whole slideshow is done currently by javascript
2008-06-08T13:24:46  <dreimark> ( i mean besides for my images, on data output the controls were probably not wanted )
2008-06-08T13:24:47  <TheSheep> cool
2008-06-08T13:48:54  <ThomasWaldmann> dreimark: http://moinmo.in/MoinMoinBugs/acl_rights_defaultDoesNotWorkForUnderlayPages   any comments to my recent idea?
2008-06-08T14:01:57  <dreimark> ThomasWaldmann: have seen it yesterday. I was not aware on the syntax given by the example. We should review the pages if there are some which needs always read rights for everyone (SystemPages).
2008-06-08T14:02:10  <dreimark> and may be we should try this on test17 first
2008-06-08T14:03:55  <dreimark> or we move this change into 1.8.
2008-06-08T14:08:19  <dreimark> bbl
2008-06-08T16:21:16  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3671:335b1d73c909 1.7/MoinMoin/script/maint/globaledit.py: modify globaledit to use better ACLs and comment on system/help pages
2008-06-08T16:26:29  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3672:457a1f43f7eb 1.7/Makefile: MoinPagesEditorGroup is not needed any more in distribution's underlay
2008-06-08T16:35:57  <ThomasWaldmann> ok, moinmo.in now runs with new underlay pages
2008-06-08T16:43:05  <zenhase> mitsuhiko: i am trying to further structure request.__init__ and request.run into a wsgi app right now
2008-06-08T16:43:22  <zenhase> mitsuhiko: and also try to think where the different contexts come into play
2008-06-08T16:43:56  <zenhase> i think much of the __init__ code can be put into lazy properties at first
2008-06-08T16:44:57  <zenhase> but i still am quite a bit confused on how those contexts should really come into being right now
2008-06-08T16:45:48  <zenhase> i guess in the wsgi app only the httpcontext will be used ... the next big part then would be the actions which are called
2008-06-08T16:46:26  <zenhase> so should those be called with say some action- or render-context already?
2008-06-08T17:11:02  <mitsuhiko> zenhase: what are those?
2008-06-08T17:13:41  <zenhase> hum?
2008-06-08T17:14:39  <zenhase> you mean those contexts?
2008-06-08T17:16:21  <dreimark> ThomasWaldmann: #acl MoinPagesEditorGroup is gone ?
2008-06-08T17:17:30  <dreimark> is AktuelleÄnderungen alocal page ?
2008-06-08T17:19:51  <dreimark> probably yes
2008-06-08T17:27:04  <dreimark> greetings from marian and a hint about a nice tool http://clonedigger.sourceforge.net/
2008-06-08T17:42:19  <dreimark> ThomasWaldmann: search on MM
2008-06-08T17:42:41  <dreimark> text search of "#acl MoinPagesEditorGroup" founds some pages
2008-06-08T17:43:08  <ThomasWaldmann> let me first clean up master17
2008-06-08T17:43:14  <dreimark> and it does highlight in the result substrings
2008-06-08T17:43:43  <dreimark> ok
2008-06-08T17:49:04  <ThomasWaldmann> man stemming :)
2008-06-08T17:52:18  <dreimark> README.migration mentiones python 2.2.2
2008-06-08T18:10:30  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3673:552982e930cc 1.7/MoinMoin/script/maint/globaledit.py: globaledit script: also cleanup *Template pages
2008-06-08T18:10:30  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3674:9fe5af566195 1.7/wiki/underlay.tar: updated underlay
2008-06-08T18:11:56  * ThomasWaldmann updates moinmo.in's underlay again
2008-06-08T18:37:10  <dreimark> action=print should not do all these span tags
2008-06-08T19:02:58  <zenhase> i was just wondering if this code for url_prefix_action in request.run does actually work oO
2008-06-08T19:03:51  <zenhase> it gets the action from the url and stores it in the var 'action' for the rest of the .run() method
2008-06-08T19:03:59  <zenhase> which never gets used it seems
2008-06-08T19:04:09  <zenhase> but still it works on moinmo.in :o
2008-06-08T19:04:15  <zenhase> how is that?
2008-06-08T19:06:52  <zenhase> oh disregard that :)
2008-06-08T19:06:58  <zenhase> it doesn't actually work ;D
2008-06-08T19:07:17  <zenhase> there is a action=... GET param too on the urls
2008-06-08T19:08:07  <ThomasWaldmann> zenhase: this stuff has to keep working like it does
2008-06-08T19:09:05  <ThomasWaldmann> (it is because of that braindead robots.txt standard)
2008-06-08T19:12:22  <zenhase> huh? could you explain that?
2008-06-08T19:12:46  <zenhase> also: it doesn't work at all
2008-06-08T19:13:10  <ThomasWaldmann> robots.txt can only exclude pathes and we don't want robots to use all our actions, therefore actions may be under some "action path"
2008-06-08T19:13:11  <zenhase> it just prefixes urls with a cute little action-prefix that has no effect whatsoever right now :)
2008-06-08T19:13:23  <ThomasWaldmann> it has
2008-06-08T19:13:31  <zenhase> well, right now it doesn't
2008-06-08T19:13:45  <zenhase> try http://moinmo.in/4ct10n/info/FlorianKrupicka/SOC2008
2008-06-08T19:13:52  <zenhase> no info-action triggered :)
2008-06-08T19:14:02  <ThomasWaldmann> that's not a url moin made
2008-06-08T19:14:17  <zenhase> well, i removed the ?action=info
2008-06-08T19:14:30  <ThomasWaldmann> you must not :)
2008-06-08T19:15:00  <zenhase> ah now i get it
2008-06-08T19:16:03  <ThomasWaldmann> (using only /action without /actionname behind it would also work)
2008-06-08T19:16:06  <zenhase> i thought it was just an fancy url thing at first
2008-06-08T19:16:31  <ThomasWaldmann> but that way, you can exclude robots easily from all actions or just from some
2008-06-08T19:16:53  <zenhase> hmm
2008-06-08T19:17:12  <zenhase> only /action without the name after it does not work
2008-06-08T19:17:34  <zenhase> it strips the page then, because it thinks it's the action name
2008-06-08T19:17:50  <ThomasWaldmann> moin doesn't guarantee to do anything useful with urls invented by someone
2008-06-08T19:18:12  * ThomasWaldmann is talking of robots.txt entries
2008-06-08T19:18:24  <zenhase> ah ok
2008-06-08T19:18:37  <zenhase> well, as said: i misunderstood what this was meant for
2008-06-08T19:19:04  <CIA-50> Byeongweon [tasyblue@gmail.com] default * 3644:f34a0146cac6 1.8-guieditor-mbyeongweon/ (2 files in 2 dirs): make fckeditor work : replace fckeditor recent version.
2008-06-08T19:19:07  <CIA-50> Byeongweon [tasyblue@gmail.com] default * 3645:c1bddba80ebd 1.8-guieditor-mbyeongweon/ (6 files in 6 dirs): solve javascript alert error
2008-06-08T19:19:49  <ThomasWaldmann> with that 3644 comment, one could think his project is finished :D
2008-06-08T19:20:23  * ThomasWaldmann doubts it :)
2008-06-08T19:23:35  * dreimark is sure in three month they are well trained in the manner of contributing 
2008-06-08T19:27:30  <zenhase> *g*
2008-06-08T19:42:29  <dreimark> ThomasWaldmann: I do want to add a lightly modified HelpOnUpdating page to the docs UPDATING.html
2008-06-08T19:42:51  <dreimark> the changes I did is about the title and the toc enumeration
2008-06-08T19:42:55  <dreimark> because of missing css
2008-06-08T19:43:36  <dreimark> s/lightly/slightly/
2008-06-08T19:47:50  <ThomasWaldmann> dreimark: see the toplevel Makefile
2008-06-08T19:49:10  <ThomasWaldmann> and please compare the page on master17 to the moinmo.in page
2008-06-08T19:56:56  <dreimark> HelpOnUpdating has no changes on MM
2008-06-08T19:57:30  <dreimark> in difference to the installing people should read that page before they start
2008-06-08T20:00:49  <dreimark> HelpOnUpdating is mentioned in Installdocs so it should be enough
2008-06-08T20:01:01  * dreimark gets invalid command name 'install-docs'
2008-06-08T20:03:28  <dreimark> ok my fault
2008-06-08T20:03:59  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3675:810eb9c0f79a 1.7/MoinMoin/search/builtin.py: xapian search results: remove strange statement disabling moin's weighting, fixes result list ordering
2008-06-08T20:03:59  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3676:79a7c1432121 1.7/MoinMoin/search/builtin.py: xapian search: clean up some code, add i18n for 'about'
2008-06-08T20:04:35  <zenhase> btw: any emacs user here?
2008-06-08T20:05:16  <zenhase> if yes: take a look at DVC, seems like a great extension to work with various distributed version control systems
2008-06-08T20:05:53  <zenhase> next to mercurial it also supports tla/arch, git, monotone and darcs is coming up
2008-06-08T20:06:36  <dreimark> ThomasWaldmann: no need for a separate HelpOnUpdating
2008-06-08T20:07:29  <ThomasWaldmann> (just wanted to make sure there is only one :)
2008-06-08T20:22:18  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3677:d28127d353fa 1.7/MoinMoin/i18n/ (42 files): update i18n
2008-06-08T20:50:15  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3678:54e654507222 1.7/docs/CHANGES: updated CHANGES
2008-06-08T20:52:24  <ThomasWaldmann> dreimark: you want to have it in rc3?
2008-06-08T20:52:53  <dreimark> wikisysnc fix - no that needs longer
2008-06-08T20:53:10  <dreimark> ThomasWaldmann:
2008-06-08T20:53:32  <ThomasWaldmann> hm? i meant helponupdating.
2008-06-08T20:54:23  <dreimark> I stopped on that because it is linked from the Installdocs page
2008-06-08T20:55:53  <ThomasWaldmann> well, looks like many people don't click that link.
2008-06-08T20:56:09  <ThomasWaldmann> (or don't read installdocs when upgrading)
2008-06-08T20:56:37  <dreimark> hmm, because they do upgrade
2008-06-08T20:57:10  <ThomasWaldmann> so some update/upgrade.html thing would help I think
2008-06-08T20:57:33  <dreimark> ok I continue and do change the Makefile too
2008-06-08T20:58:58  <zenhase> got a question: does hg.moinmo.in use hg-ssh for repo-access?
2008-06-08T20:59:30  <zenhase> and how secure is such a setup against shell access if you want to deploy it in a say less trusted group of contributers?
2008-06-08T21:04:53  <dreimark> 1:0
2008-06-08T21:06:26  <ThomasWaldmann> zenhase: there is no direct shell login
2008-06-08T21:07:15  <ThomasWaldmann> of course there could be some strange bug in hg-ssh maybe making it possible to execute unpriviledged commands
2008-06-08T21:07:34  <ThomasWaldmann> (but that risk you have with every webapp also)
2008-06-08T21:07:39  <CIA-50> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3679:c19b7a3d7539 1.7/Makefile: Makefile: added fetching UPDATE instructions
2008-06-08T21:07:41  <CIA-50> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3680:045a9528f095 1.7/ (44 files in 3 dirs): merged main
2008-06-08T21:08:30  <ThomasWaldmann> for hg-ssh you additionally require a installed ssh pubkey, so abuse should be rare
2008-06-08T21:08:50  <zenhase> i also just found contrib/hgsh/hgsh.c in mercurial
2008-06-08T21:08:56  <zenhase> seems like an alternative
2008-06-08T21:09:22  <dreimark> hg diff
2008-06-08T21:09:28  <dreimark> wrong window
2008-06-08T21:10:59  <ThomasWaldmann> zenhase: btw, some form.get support would be nice in the new request class
2008-06-08T21:11:51  <ThomasWaldmann> (for the usual stuff we need: get unicode, get int, get boolean maybe)
2008-06-08T21:52:06  <zenhase> you mean like get it decoded?
2008-06-08T21:52:15  <zenhase> erm
2008-06-08T21:52:28  <zenhase> even more than decoded ... get it coerced
2008-06-08T21:52:54  <zenhase> i would not really put that into the request object
2008-06-08T21:53:15  <zenhase> perhaps something similar to formencode
2008-06-08T21:53:57  <zenhase> i will note that down, some simple form2python facility
2008-06-08T21:54:43  <zenhase> i think somethings similar to formencode can be cooked up in a few lines of code, but i wouldn't make it a priority for the gsoc-project
2008-06-08T21:55:03  <zenhase> (with meaning some 100 here)
2008-06-08T22:01:46  <dreimark> ThomasWaldmann: test_search +56 fails for me
2008-06-08T22:02:04  * dreimark seems to have a fix for the wikisync problem
2008-06-08T22:04:19  <dreimark> xorAxAx: http://moinmo.in/MoinMoinBugs/WikiSyncMissingInterWikiNameOnRemoteSide
2008-06-08T22:04:36  <dreimark> do you have a better solution?
2008-06-08T22:04:41  <dreimark> xorAxAx:
2008-06-08T22:10:29  <xorAxAx> dreimark: i dont see how that should help
2008-06-08T22:10:58  <xorAxAx> dreimark: because the list has at least two elements ... how should it be not nonzero to reach your "or ''"?
2008-06-08T22:12:47  <dreimark> it does
2008-06-08T22:13:17  <xorAxAx> how?
2008-06-08T22:13:17  <dreimark> if remote wiki name is None packline crashes
2008-06-08T22:13:50  <dreimark> sorry
2008-06-08T22:13:52  <xorAxAx> umm
2008-06-08T22:14:01  <dreimark> remote.get_interwiki_name() is None
2008-06-08T22:14:10  <dreimark> if you don't enter an interwikiname
2008-06-08T22:14:12  <xorAxAx> your patch is still broken
2008-06-08T22:14:26  <xorAxAx> but at least i know the correct line where it crashed ..
2008-06-08T22:14:52  <dreimark> \o/
2008-06-08T22:15:19  * dreimark waits for the real fix
2008-06-08T22:15:54  <xorAxAx> the correct fix is to use         remote_full_iwid = remote.iwid_full
2008-06-08T22:16:46  <xorAxAx> and the allow_none hunk
2008-06-08T22:22:35  <ThomasWaldmann> hmm, that test breakage is due to my fix for "negative only" search terms
2008-06-08T22:28:35  * zenhase is wondering a bit about find_remote_addr
2008-06-08T22:29:12  <zenhase> proxies_trusted does not seem to be configurable via normal means
2008-06-08T22:29:18  <zenhase> (meaning wikiconfig)
2008-06-08T22:29:29  <ThomasWaldmann> yes
2008-06-08T22:29:31  <zenhase> so is this actually used in setups this way?
2008-06-08T22:29:38  <ThomasWaldmann> you should fix that somehow :)
2008-06-08T22:29:41  <zenhase> with modified Moin source?
2008-06-08T22:30:13  <CIA-50> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3681:a1a9ac307c0e 1.7/MoinMoin/ (action/SyncPages.py xmlrpc/__init__.py): SyncPages, xmlrpc.process: bug fix for WikiSyncMissingInterWikiNameOnRemoteSide (thanks to Alexander Schremmer)
2008-06-08T22:30:13  <ThomasWaldmann> it was a "better edit source than no proxy support at all" thing
2008-06-08T22:31:14  <zenhase> ah ok
2008-06-08T22:31:33  <zenhase> well, no problem putting that right
2008-06-08T22:31:59  <zenhase> which reminds me of something i wanted to talk about
2008-06-08T22:32:20  <zenhase> what is your stance on threadlocal variables?
2008-06-08T22:32:48  <zenhase> 'your' meaning all developers in this case
2008-06-08T22:33:29  <zenhase> because i always wondered about config being always tighted to the request object
2008-06-08T22:33:53  <zenhase> it's not that i want to change that right now
2008-06-08T22:34:07  <zenhase> it was just something that's in the back of my head
2008-06-08T22:34:30  <ThomasWaldmann> for now, you have to be py 2.3 compatible
2008-06-08T22:34:47  <zenhase> isn't there threadlocalism in py2.3?
2008-06-08T22:35:35  <ThomasWaldmann> and what has config to do with thread local?
2008-06-08T22:36:41  <zenhase> it's more like an application global setting
2008-06-08T22:37:06  <zenhase> but still request has to get passed into the last corner most of the time, often just because of some settings in the cfg
2008-06-08T22:37:54  <zenhase> i do realize, that there is the farmconfig problem, which needs to get the config-object according to the wiki-instance visited
2008-06-08T22:40:20  <dreimark> ThomasWaldmann: do you have got new keys from marian too?
2008-06-08T22:41:39  <ThomasWaldmann> dreimark: not iirc
2008-06-08T22:42:18  <ThomasWaldmann> zenhase: config is not global and not bound to a thread
2008-06-08T22:43:47  * dreimark forwards the mail
2008-06-08T22:44:31  <dreimark> hmm there was a cc  to you
2008-06-08T22:44:43  <dreimark> let's see if I have more luck
2008-06-08T22:45:11  <ThomasWaldmann> i have his mail
2008-06-08T22:49:26  <zenhase> ThomasWaldmann: it was an idea of how config could be accessed apart from request-objects, not the status quo
2008-06-08T22:49:30  <dreimark> ok
2008-06-08T22:50:16  <zenhase> ThomasWaldmann: i've seen it in other applications and wanted to know what moin-devs think about this approach
2008-06-08T22:52:47  <ThomasWaldmann> are you sure you want to turn that all upside down?
2008-06-08T22:53:08  <ThomasWaldmann> my impression is that you are touching too much stuff
2008-06-08T22:54:55  <zenhase> ah, i already said that i don't want to touch that now
2008-06-08T22:55:15  <zenhase> it was something i wanted to ask opinion about
2008-06-08T22:56:58  <ThomasWaldmann> there could be a site manager object keeping configuration and other stuff for all sites in the farm
2008-06-08T22:58:09  <zenhase> how would any code needing access to configuration access it's particular config?
2008-06-08T22:58:58  <xorAxAx> zenhase: have you asked your mentor already?
2008-06-08T23:01:01  <zenhase> it was an open question to everyone here :)
2008-06-08T23:01:24  <zenhase> erm
2008-06-08T23:01:40  <zenhase> s/open/overt
2008-06-08T23:06:37  <xorAxAx> zenhase: but your mentor was selected to give you guidance and i dont see him involved
2008-06-08T23:07:14  <xorAxAx> .oO(why dont you call the objects c and g and call it pylonicrequest *duck*)
2008-06-08T23:09:15  <zenhase> because it wasn't meant as a question particular to the success of my gsoc task
2008-06-08T23:09:34  <zenhase> but something that was bothering me in general
2008-06-08T23:10:48  <xorAxAx> regarding the question: note that there are wsgi compliant servers that might fail if you use thread-local vars
2008-06-08T23:11:48  <xorAxAx> imagine a wsgi server benefitting from the app being a generator and running them in different green threads
2008-06-08T23:12:12  <xorAxAx> but indeed, thats a very hypothetical scenario
2008-06-08T23:12:49  <xorAxAx> i am personally fine with the tls naming pattern
2008-06-08T23:13:16  <xorAxAx> what do you, mitsuhiko, mentor of the bunny, say?
2008-06-08T23:20:41  <mitsuhiko> reading backlog
2008-06-08T23:21:24  <ThomasWaldmann> johill: can you please have a look at the search.queryparser?
2008-06-08T23:22:20  <ThomasWaldmann> i did a partial fix be removing one of the "only one item in and expression" optimizations
2008-06-08T23:22:51  <ThomasWaldmann> i guess removing the other same kind optimization a bit upwards in the code would completely fix it,
2008-06-08T23:23:10  <ThomasWaldmann> but it breaks the tests and generates rather deeply nested expressions.
2008-06-08T23:23:25  <ThomasWaldmann> so I am not sure this is the best way to deal with it.
2008-06-08T23:26:28  <zenhase> xorAxAx: this greenthreads scenario is something mitsuhiko already solved in werkzeug
2008-06-08T23:26:54  <xorAxAx> zenhase: yeah, it can be solved by cooperation of course ... but not generally
2008-06-08T23:26:58  <zenhase> only version i don't know about how to solve it there is an async server like twisted
2008-06-08T23:30:00  <zenhase> wb mitsuhiko btw :)
2008-06-08T23:30:14  <mitsuhiko> hi zenhase :)
2008-06-08T23:30:58  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3682:a5e33ae8f469 1.7/MoinMoin/search/queryparser.py: fix xapian search for '-a or b' query (see also 3669:0cbd8b41aebd), tests broken
2008-06-08T23:30:59  <CIA-50> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3683:02a602f955f6 1.7/ (3 files in 3 dirs): merged main
2008-06-08T23:31:01  <zenhase> mitsuhiko: i am currently structuring for the context object approach
2008-06-08T23:31:17  <zenhase> mitsuhiko: right now i would have following context-objects:
2008-06-08T23:31:21  <mitsuhiko> xorAxAx: i don't see how i should answer that question :)
2008-06-08T23:31:26  <zenhase> HTTPContext - reuqest on the lowest layer
2008-06-08T23:31:32  <ThomasWaldmann> johill: see 3682:a5e33ae8f469 and 3669:0cbd8b41aebd
2008-06-08T23:31:51  <zenhase> RenderContext - perhaps bad name, but is the context that should get passed to actions (and therefore sendPage)
2008-06-08T23:31:59  <xorAxAx> mitsuhiko: well, everybody has one reply for free :)
2008-06-08T23:32:14  <zenhase> XMLRPCContext - factored out xml-rpc into an extra wsgi app
2008-06-08T23:34:16  <dreimark> good night
2008-06-08T23:34:19  <zenhase> right now it's a bit hard to separate responsibility in those contexts, i guess RenderContext will end up having some references to HTTPContext (and therefore ultimately the werkzeug-request underlying it) in the beginning
2008-06-08T23:34:45  <mitsuhiko> i'm afraid it will
2008-06-08T23:35:06  <zenhase> even deep in the actions and macros there is still code that accesses request-level variables
2008-06-08T23:35:26  <zenhase> (i don't count cfg in on this, this will be present on every context)
2008-06-08T23:35:36  <zenhase> (has too right now ...)
2008-06-08T23:35:42  <zenhase> s/too/to
2008-06-08T23:35:58  <zenhase> which is why i did pose this question btw. :)
2008-06-08T23:36:43  <zenhase> but it's not that horrible for the time being, so it was just something to consider for later on
2008-06-08T23:37:39  <zenhase> ah, also i have another topic on my notebook here:
2008-06-08T23:38:05  <zenhase> it says session handling
2008-06-08T23:39:48  <zenhase> mitsuhiko: iirc you weren't very keen on sessionmiddlewares too, right? so should stay session-handling where it is right now and just setup for some .session object on request in the setup phase like it's done in .__init__ and .run now?
2008-06-08T23:40:16  <zenhase> or is there another idea?
2008-06-08T23:51:32  <zenhase> i'll afk a bit