MoinMoin: MoinMoinChat/Logs/moin-dev/2009-02-03

2009-02-03T00:00:46  <Noya> and I don't think I will apply as a student again... it really was fun, and I learned really really much
2009-02-03T00:01:10  <Noya> but it is also much much work :)
2009-02-03T00:01:33  <dreimark> that's true but mentoring can be much work too
2009-02-03T00:01:40  <Noya> no doubt
2009-02-03T00:01:50  <Noya> but hopefully not as much as being a student ;)
2009-02-03T00:02:00  <dreimark> sure
2009-02-03T00:02:12  <Noya> dreimark: but today I already submitted two patches... no need for gsoc if I can help a bit right now :p
2009-02-03T00:02:24  <dreimark> very good
2009-02-03T00:03:54  <Noya> seriously get this api-documentation to the frontpage... this is really helpful
2009-02-03T00:04:40  <dreimark> I know but it should be engl. beforehand
2009-02-03T00:05:53  <dreimark> TheSheep: has already a documentation started about the theming stuff and it would be very nice to get both into some kind of moin book
2009-02-03T00:08:21  <Noya> dreimark: and I always thought the web is full of translators :)
2009-02-03T00:08:28  <dreimark> after 2.0 we may be have time or we get scholar student at next ghop
2009-02-03T00:08:44  <dreimark> Noya: it is but not at the time you want one
2009-02-03T00:08:54  <Noya> ;)
2009-02-03T00:09:02  <dreimark> google/translate_t did a great job
2009-02-03T02:23:00  <grzywacz> Noya, does your `send_user_created_text` patch work for you? ;)
2009-02-03T02:25:09  <CIA-38> Karol 'grzywacz' Nowak <grywacz@gmail.com> default * 4267:9a3e85b44c25 1.8/jabberbot/ (commands.py xmppbot.py): Applying a patch from Noya to use an RFC compliant message type.
2009-02-03T02:25:58  <Noya> grzywacz: sure :)
2009-02-03T02:26:06  <Noya> grzywacz: doesn't it for you?
2009-02-03T02:26:49  <grzywacz> Noya, it looks good, but I'm going to sleep now and don't feel like launching everything to test ;)
2009-02-03T02:27:07  <Noya> as I stated in the bugreport I tested it in my own wiki :) for about a week now
2009-02-03T02:27:23  <Noya> grzywacz: yeah, just try it when you've more time :)
2009-02-03T02:27:50  <grzywacz> Must have missed that clarification. :)
2009-02-03T02:30:40  <CIA-38> Karol 'grzywacz' Nowak <grywacz@gmail.com> default * 4268:544679a8a444 1.8/jabberbot/xmppbot.py: Commiting a patch from Noya to fix user creation notifications.
2009-02-03T02:31:00  <grzywacz> I will merge them to 1.9 tomorrow...
2009-02-03T02:31:03  <grzywacz> Noya, thanks again. :)
2009-02-03T02:32:57  <Noya> grzywacz: no problem, thank you for merging ;)
2009-02-03T02:33:27  <Noya> grzywacz: please do me a favour and test the patch when you have time, just in case :p
2009-02-03T02:33:32  <Noya> (but it should be fine, don't worry)
2009-02-03T02:33:37  <grzywacz> Sure. Tomorrow.
2009-02-03T02:33:56  <Noya> grzywacz: and if you have time too you could implement the send-as-from-message  thing too ;)
2009-02-03T02:34:00  <Noya> grzywacz: so, now I am tired
2009-02-03T02:34:03  <Noya> need some sleep
2009-02-03T02:34:07  <Noya> see you then :)
2009-02-03T03:07:55  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4532:65c0426f2b1f 1.9/MoinMoin/web/ (flup_frontend.py frontend.py): flup cgi/fcgi adaptor: use None as interface default
2009-02-03T03:10:51  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4533:58ceb1d1c38d 1.9/wiki/server/moin.cgi: add a workaround for CGI autodetection problem in flup <= 1.0.1
2009-02-03T03:49:00  * ThomasWaldmann plays with apache / cgi / basic auth
2009-02-03T05:43:16  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4534:0b176807b07c 1.9/MoinMoin/wsgiapp.py: bugfix: create dummy user obj at the right place
2009-02-03T05:43:17  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4535:0cac33a38cb9 1.9/MoinMoin/support/werkzeug/wrappers.py: werkzeug: fix remote_user value
2009-02-03T05:43:18  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4536:9b4c9f73b632 1.9/MoinMoin/auth/http.py: make http basic auth work (tested with Apache2/CGI)
2009-02-03T10:51:55  <dreimark> moin
2009-02-03T12:52:43  <ronny> ThomasWaldmann: how wide is the feature of nested pages used?
2009-02-03T12:53:12  <ThomasWaldmann> ronny: you mean subpages?
2009-02-03T12:53:13  <ronny> (its a bit tricky to map those to branching patterns
2009-02-03T12:53:18  <ronny> yeah
2009-02-03T12:53:30  <ThomasWaldmann> widely used.
2009-02-03T12:53:49  <ronny> im just thinking about ideas to make merges within wikis mappable to dvcs's
2009-02-03T12:53:52  <ThomasWaldmann> and due to the way they are implemented, there are not much related problems
2009-02-03T12:54:20  <ronny> the main issue is dvcs's are whole tree granularity and wikis are page-granularity
2009-02-03T12:55:07  <ThomasWaldmann> TheSheep and pawel pacana know more about this topic
2009-02-03T12:56:00  <ronny> also i'll need to tinker a bit more with anyvc
2009-02-03T12:56:16  <ThomasWaldmann> (subpages are just pages that happen to have a / in their name, btw)
2009-02-03T12:56:17  <ronny> no history support atm, as well as no unconnected commits
2009-02-03T12:56:50  <ronny> ThomasWaldmann: oh, that makes things more simple
2009-02-03T12:57:07  <ronny> (ie hierachy is emulated by spliting a name instead of having an actual hierachy
2009-02-03T12:57:17  <ThomasWaldmann> yes
2009-02-03T12:57:24  <ronny> fits hg perfectly
2009-02-03T12:57:38  <ronny> is a bit weird for git/bzr ( they have trees)
2009-02-03T12:57:41  <ThomasWaldmann> it does?
2009-02-03T12:58:06  <ronny> hg tracks files by their path, there are no trees in the manifest
2009-02-03T13:01:17  <ThomasWaldmann> ok
2009-02-03T13:02:48  <ronny> the issue is the granulatity
2009-02-03T13:02:57  <ronny> whitzbit could be more usefull later tho
2009-02-03T13:03:15  <ronny> (its a gnome project to deal with file granulatiry data sync betwen desktops)
2009-02-03T13:03:26  <ThomasWaldmann> btw, note to all: i think we should rename auth.http.HTTPAuth
2009-02-03T13:03:45  <ronny> bbl, getting food
2009-02-03T13:04:40  <ThomasWaldmann> (because it has nothing to do with http auth, but just uses REMOTE_USER env var, no matter what auth method was used by the web server to authenticate that user)
2009-02-03T13:05:11  <ThomasWaldmann> so, how do we call that?
2009-02-03T13:05:26  <ThomasWaldmann> RemoteUserAuth?
2009-02-03T13:14:14  <ThomasWaldmann> or do we want that configurable, like EnvVarAuth(name='REMOTE_USER') ?
2009-02-03T13:21:52  <TheSheep> those names are confusing
2009-02-03T13:25:17  <Noya> good morning :)
2009-02-03T13:26:07  <ThomasWaldmann> TheSheep: that's why I am asking (but at least, they are not wrong)
2009-02-03T13:27:27  <TheSheep> ExternalAuth? hmmm....
2009-02-03T13:27:32  <ThomasWaldmann> some people already wondered about why our "http auth" does not do http auth on its own, because they were unaware that they need apache doing that
2009-02-03T13:27:35  <TheSheep> a little too vague
2009-02-03T13:28:27  <TheSheep> PassAuth ?
2009-02-03T13:28:40  <TheSheep> from WSGI's "WSGIPassAuthorization" variable
2009-02-03T13:28:56  <ThomasWaldmann> with configurable env var name one could maybe do nice hacks
2009-02-03T13:29:03  <TheSheep> EnvironAuth sounds best so far, I think
2009-02-03T13:29:19  <TheSheep> but requires knowledge of internal workings
2009-02-03T13:29:44  <TheSheep> UpstreamAuth?
2009-02-03T13:29:58  <TheSheep> PassedAuth?
2009-02-03T13:30:03  <ThomasWaldmann> WeBelieveItAuth
2009-02-03T13:30:12  <TheSheep> InheritedAuth?
2009-02-03T13:30:48  <TheSheep> something that tells you the auth is done a step higher...
2009-02-03T13:31:00  <TheSheep> not how it's passed
2009-02-03T13:31:09  <ThomasWaldmann> GivenAuth
2009-02-03T13:32:03  <ThomasWaldmann> ReUseAuth
2009-02-03T13:32:05  <TheSheep> AlienAuth
2009-02-03T13:32:20  <TheSheep> IWantToBelieveAuth ;)
2009-02-03T13:32:27  <ThomasWaldmann> Alien2Auth
2009-02-03T13:32:41  <TheSheep> AlienVsPredatorAuth
2009-02-03T13:33:16  <TheSheep> BigBrotherAuth
2009-02-03T13:34:16  <TheSheep> I like Given and Reuse best
2009-02-03T13:34:30  <ThomasWaldmann> btw, I have seen some dummyauth recently. that can be done via environment, no need for extra plugin.
2009-02-03T13:34:53  <TheSheep> it was a quick hack
2009-02-03T13:34:57  <TheSheep> it doesn't work
2009-02-03T13:35:06  <TheSheep> somehow the settings are not saved
2009-02-03T13:35:16  <TheSheep> no idea why
2009-02-03T13:35:22  <TheSheep> I will have to remove it
2009-02-03T13:35:33  <TheSheep> or investigate
2009-02-03T13:36:15  <ThomasWaldmann> it could be GivenAuth(env_var='REMOTE_USER', user_name=None)
2009-02-03T13:37:21  <ThomasWaldmann> or better both =None and using REMOTE_USER if nothing is given
2009-02-03T13:38:34  <ThomasWaldmann> i guess that sounds like a plan :)
2009-02-03T13:40:29  <TheSheep> I hope the rename is schduled for moin 2.0?
2009-02-03T13:41:32  <ThomasWaldmann> 1.9
2009-02-03T13:41:45  <TheSheep> argh
2009-02-03T13:42:35  <ThomasWaldmann> I'll write a HINT about it.
2009-02-03T13:43:39  <ThomasWaldmann> 2.0 will be the release when we break storage, so the auth change rather fits into 1.9
2009-02-03T13:44:33  <TheSheep> MoinMoin: what do you want do break today? ;)
2009-02-03T13:45:39  <ThomasWaldmann> yeah, we could be still pikipiki compatible
2009-02-03T13:45:55  * TheSheep despairs at the loss of oportunity
2009-02-03T13:48:04  <ThomasWaldmann> we could keep http.HTTPAuth for 1.9 as a redirect that emits a deprecation warning
2009-02-03T13:49:40  <TheSheep> sounds good
2009-02-03T13:49:43  <ThomasWaldmann> and I guess I'll move the code for that to __init__, because it is a rather fundamental thing
2009-02-03T13:50:04  <TheSheep> I was even wondering about just enabling it by default
2009-02-03T13:50:18  <ThomasWaldmann> so one can do from MoinMoin.auth import GivenAuth
2009-02-03T13:50:41  <TheSheep> But I'm not sure how different servers handle this, so there are some security considerations
2009-02-03T13:50:55  <ThomasWaldmann> that could be unwanted in some cases
2009-02-03T13:51:11  <TheSheep> for example, if there is no http auth set up, but I pass user and password in the url, will the REMOTE_USER get set?
2009-02-03T13:51:56  <TheSheep> it doesn't in Apache and LIghty, but no idea about other servers
2009-02-03T13:52:07  <ThomasWaldmann> it can't verify the password, so I hope it doesn't set R_U
2009-02-03T14:16:34  <dreimark> moin
2009-02-03T14:17:09  <dreimark> do we need to have the name Auth in that name?
2009-02-03T14:21:21  <ThomasWaldmann> it is there for the others, too
2009-02-03T14:35:27  <TheSheep> RemoteUserAuth
2009-02-03T14:35:29  <TheSheep> :)
2009-02-03T15:04:36  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4537:f8bf8de778f2 1.9/MoinMoin/ (auth/__init__.py auth/http.py config/multiconfig.py):
2009-02-03T15:04:36  <CIA-38> move auth.http.HTTPAuth to auth.GivenAuth, see details below
2009-02-03T15:04:36  <CIA-38> old name was not reflecting what it really did (just use REMOTE_USER env var)
2009-02-03T15:04:36  <CIA-38> compatibility and deprecation warning code added as auth.http.HTTPAuth
2009-02-03T15:04:36  <CIA-38> extended GivenAuth so env_var can give another env. var. name
2009-02-03T15:04:40  <CIA-38> extended GivenAuth so a fixed user_name can be given that is considered authenticated
2009-02-03T15:04:42  <CIA-38> trusted_auth_methods default now includes 'http' and 'given' for compatibility
2009-02-03T15:23:24  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4269:776cc7f6f6c6 1.8/MoinMoin/userprefs/oid.py: compatibility fix for python-openid 2.x.x (also works with 1.x.x), fixes crash when trying to associate moin user to openid
2009-02-03T15:54:30  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 345:630b3cfb01dc 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: macro to collect data from definition lists from subpages pages into a databrowser widget table.
2009-02-03T16:51:36  <ThomasWaldmann> hmm, looks like zenhase has killed that special treatment for ntlm/sspi auth
2009-02-03T16:52:22  <ThomasWaldmann> (stripping windows domain prefix, ...)
2009-02-03T17:15:47  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 346:c8f59d078d98 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: support more than only one level
2009-02-03T17:15:48  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 347:a701beab4b9c 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: empty cells fixed
2009-02-03T17:15:54  <TheSheep> that's what I hate in rewrites
2009-02-03T17:24:49  <ThomasWaldmann> hm?
2009-02-03T17:25:11  <ThomasWaldmann> ah. I am just readding some code for that.
2009-02-03T17:26:50  <ThomasWaldmann> TheSheep: do you have experience with http auth username schemes in bigger environments?
2009-02-03T17:27:07  <ThomasWaldmann> ok, there is that DOMAIN\username thing, treated
2009-02-03T17:27:26  <ThomasWaldmann> Joe Doe -> JoeDoe, treated
2009-02-03T17:27:41  <ThomasWaldmann> joe doe -> Joe Doe, treated
2009-02-03T17:27:58  <ThomasWaldmann> anything else, emailish maybe?
2009-02-03T17:27:59  <TheSheep> no, I don't have any experience in corporate environments
2009-02-03T17:28:19  <TheSheep> joe.doe, joe_doe, joe-doe
2009-02-03T17:28:41  <TheSheep> joe+doe
2009-02-03T17:28:53  <TheSheep> joe%20doe
2009-02-03T17:29:40  <ThomasWaldmann> JoeDoe@company.com ?
2009-02-03T17:30:17  <TheSheep> I suppose cutting off everything after @ and < would be a good idea
2009-02-03T17:31:02  <TheSheep> then again, I soubt anybody would really use such a username
2009-02-03T17:33:59  <ThomasWaldmann> i only want to treat the most common transformation cases by flags
2009-02-03T17:34:18  <ThomasWaldmann> everybody needed something more advanced can add a transformation function
2009-02-03T17:34:30  <ThomasWaldmann> ..needing..
2009-02-03T17:42:52  <ThomasWaldmann> TheSheep: do you know what encoding REMOTE_USER is?
2009-02-03T17:43:07  <TheSheep> ascii
2009-02-03T17:43:35  <TheSheep> url-encoded
2009-02-03T17:43:42  <TheSheep> %-encoded, I mean
2009-02-03T17:51:49  <ThomasWaldmann> hmm, htpasswd seems to dislike umlauts
2009-02-03T17:52:25  <ThomasWaldmann> it writes the entry into the file, but it doesn't work with basic auth afterwards
2009-02-03T17:52:37  <TheSheep> I'm not a fan either ;)
2009-02-03T17:52:46  <TheSheep> and sharp-s looks like beta to me
2009-02-03T17:55:37  <ThomasWaldmann> TheSheep: i just get "Thomas Waldmann" (str), no % stuff
2009-02-03T17:56:03  <TheSheep> :/
2009-02-03T17:56:26  <ThomasWaldmann> maybe werkzeug's request.remote_user is preprocessed somehow?
2009-02-03T17:56:36  <TheSheep> that's possible
2009-02-03T18:01:50  <ThomasWaldmann> TheSheep: my apache uses latin1 encoding for the username :|
2009-02-03T18:02:12  <ThomasWaldmann> but my system uses utf-8 otherwise
2009-02-03T18:02:36  <TheSheep> ThomasWaldmann: what if you use something like łąka@foo.bar
2009-02-03T18:03:20  <ThomasWaldmann> i had trouble getting that ü into htpasswd file, don't ask about such weird stuff
2009-02-03T18:04:34  * TheSheep obediently stops asking
2009-02-03T18:04:36  <ThomasWaldmann> but it somehow sucks if the htpasswd tool and the apache server live in different worlds
2009-02-03T18:05:07  <ThomasWaldmann> one speaking utf8 and the other expecting latin1
2009-02-03T18:05:29  <TheSheep> I've given up on apache long time ago
2009-02-03T18:06:54  <TheSheep> not that lighty is any bettter, but at least simplier
2009-02-03T18:11:14  <ThomasWaldmann> ok, wikiutil.decodeUnknownInput does it for now
2009-02-03T19:10:09  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4538:d0afc869ab36 1.9/MoinMoin/auth/__init__.py: add some decoding/transformations to GivenAuth (similar to what we had hardcoded in 1.8)
2009-02-03T19:18:20  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 348:d3547d33dd32 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: filter_pattern introduced to select which rows you want in the result
2009-02-03T19:18:22  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 349:33cb55aaa955 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: keyword align introduced
2009-02-03T19:18:23  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 350:2eda7e0194e5 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: flagged not to use transpose with filter_selection
2009-02-03T19:18:24  <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 351:402214f6607c 1.7-extensions/data/plugin/macro/CollectLists.py: CollectLists: flagged not to use transpose with filter_pattern. removing of a column isn't implemented yet.
2009-02-03T20:16:05  <ThomasWaldmann> grzywacz: how's the jabber auth stuff going?
2009-02-03T20:16:25  <ThomasWaldmann> i guess except that we have no known bigger bugs left, right?
2009-02-03T20:17:46  <grzywacz> ThomasWaldmann, slowly, auth in 1.8 is weird and I was busy with a couple of other things.
2009-02-03T20:18:17  <grzywacz> ThomasWaldmann, basically I figured out I need to write an auth method that let's user in if it has a jid set and is valid. :P
2009-02-03T20:18:29  <grzywacz> ThomasWaldmann, correct me if there's a better way to do that.
2009-02-03T20:22:20  <ThomasWaldmann> no idea. the jabber code is from yours and the auth stuff is from johill.
2009-02-03T20:22:52  <ThomasWaldmann> and in 1.9 (what I've fixed recently), it seems to be different.
2009-02-03T21:22:58  <grzywacz> ThomasWaldmann, jabber code is irrelevant. All I need is to get an auth token registered without any authentication.
2009-02-03T21:27:07  <ThomasWaldmann> for 1.9 I just used the same value as in the session cookie
2009-02-03T21:49:40  <dreimark> re
2009-02-03T23:25:57  * dreimark found a bug in databrowser widget related to make_unique_id