MoinMoin: MoinMoinChat/Logs/moin-dev/2007-06-19

2007-06-19T00:07:30  <neagulm> dreimark, I'm going to sleep. I have school from 8 EET
2007-06-19T00:07:58  <dreimark> neagulm: good night
2007-06-19T00:08:05  <neagulm> good night
2007-06-19T00:20:00  <dreimark> good night
2007-06-19T01:31:30  <CIA-27> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> * 2165:eed1c7da25eb 1.7/MoinMoin/ (_tests/test_user.py user.py): test_user: fix wrong subscription tests and a few other problems
2007-06-19T01:43:10  <zenhase> hmm, i am still structuring my thoughts
2007-06-19T01:44:05  <zenhase> i will catch some sleep now but tomorrow i would like to talk about the design ideas with you guys
2007-06-19T01:44:19  <zenhase> cu tomorrow
2007-06-19T08:39:29  <ThomasWaldmann> moin
2007-06-19T09:22:27  <grzywacz> moin
2007-06-19T09:22:30  * grzywacz back to sleep -_-;
2007-06-19T09:58:26  <zenhase> moin
2007-06-19T10:13:22  <ffesti> moin
2007-06-19T12:06:48  <dreimark> moin
2007-06-19T12:25:52  <ThomasWaldmann> dreimark: moin. I refactored some tests a bit yesterday. But that testTooMuchWantedArguments still crashes.
2007-06-19T12:26:39  <ThomasWaldmann> (or to be more exact: the code it tests crashes)
2007-06-19T13:21:03  <johill> grzywacz: is the 'secret' supposed to be used for more than that bot? otherwise imho you should rename it. Also, you should somehow disallow using the default secret
2007-06-19T13:21:22  <johill> grzywacz: maybe use something like secret=None and then check for secret and raise an ecxeption where it's required
2007-06-19T13:31:46  <dreimark> ThomasWaldmann: that part of the code of the argParser is not finished. IMHO it should use the two known arguments and ignore or probably add the extra argument into another dict to inherit it on an other routine. Currently it crashes if some one gives more arguments as defined.
2007-06-19T13:32:23  <dreimark> ThomasWaldmann: I thought about how we can add spam to neagulms testwiki
2007-06-19T13:35:16  <dreimark> one possibility is to enter pages by this mail to page extension you added. So I could sent some of the mails my junkcontrol has recognizied as spam already. Can you give a hint where to look at MoinMaster
2007-06-19T13:53:50  <grzywacz> johill, wise advice.
2007-06-19T14:13:42  <ThomasWaldmann> johill: i have thought about secrets = { 'key': 'value', ...}
2007-06-19T14:15:04  <ThomasWaldmann> dreimark: you can find lots of reverted spam on mm site
2007-06-19T15:40:42  <lanius> xorAxAx: ThomasWaldmann: can you recommend any alternative for hgmerge for windows?
2007-06-19T15:40:50  <xorAxAx> lanius: hmm?
2007-06-19T15:40:57  <xorAxAx> lanius: which merge program are you using?
2007-06-19T15:41:03  <lanius> no one yet ;)
2007-06-19T15:41:06  <xorAxAx> lanius: did you install and configure kdiff3 as i said?
2007-06-19T15:41:10  <lanius> because hgmerge is not avialable
2007-06-19T15:41:11  <xorAxAx> how do you want to merge then ....
2007-06-19T15:41:14  <xorAxAx> oh well
2007-06-19T15:41:22  <ThomasWaldmann>  i dont use any :)
2007-06-19T15:41:25  <lanius> that's why i ask, ok kdiff then
2007-06-19T15:41:39  <xorAxAx> lanius: and you can google for the correct hgmerge.cmd as i said
2007-06-19T15:41:42  <xorAxAx> that will call kdiff3
2007-06-19T15:41:48  <lanius> yes that's what i found out
2007-06-19T15:42:15  <xorAxAx> and then there is just a single problem - kdiff3 only accepts a fixed line ending
2007-06-19T15:42:25  <lanius> that means
2007-06-19T15:42:26  <xorAxAx> depending on how you configured your hg, you must set it to UNIX or DOS
2007-06-19T15:42:43  <xorAxAx> it means that i t cannot auto detect wether it should use one of those line endings
2007-06-19T15:45:57  <johill> grzywacz: :) I don't think secrets = {...} is all that useful, although then we could have a common "please set secret first" mechanism
2007-06-19T15:46:45  <ThomasWaldmann> well, it is useful when you need more than 1 secret
2007-06-19T15:46:51  <lanius> xorAxAx: TortoiseMege seems to work good, already hat that installed
2007-06-19T15:46:56  <ThomasWaldmann> and we already have one for mailimport
2007-06-19T15:47:01  <xorAxAx> lanius: also ok
2007-06-19T15:49:14  <lanius> though i don't now why it opens file to merge which contain no conflicts
2007-06-19T15:49:27  <xorAxAx> hehe :)
2007-06-19T15:49:47  <xorAxAx> yes, it needs to support to "go away" automatically in such cases
2007-06-19T15:49:53  <xorAxAx> kdiff3 does that
2007-06-19T16:19:28  <dreimark> bbl
2007-06-19T16:32:02  <johill> ThomasWaldmann: why do we need multiple secrets anyway?
2007-06-19T16:32:21  <johill> afaict with all of them you have write access to the wiki
2007-06-19T16:32:31  <johill> s/all/any/
2007-06-19T18:27:39  <lanius> ThomasWaldmann: seems like moin-1.7 has some new bugs
2007-06-19T18:27:59  <lanius> ThomasWaldmann: saving users does not work neither [[SystemAdmin]]
2007-06-19T18:56:12  <ThomasWaldmann> lanius: what happens?
2007-06-19T19:03:03  <Fabi> zenhase, still waiting for your design document...
2007-06-19T19:03:34  <Fabi> doesn't need to be that epic
2007-06-19T19:07:22  <ThomasWaldmann> how about a design wikipage :D
2007-06-19T19:07:59  <Fabi> I think that was the plan...
2007-06-19T19:12:13  <xorAxAx> grzywacz: _subscribable_events -- s/b/p/
2007-06-19T19:13:06  <xorAxAx> grzywacz:  get_subscribable_events() should not use this caching pattern but generate the list at module loading time
2007-06-19T19:13:11  <xorAxAx> looks saner
2007-06-19T19:13:47  <xorAxAx> grzywacz: oh, its indeed with a b and not a p
2007-06-19T19:15:10  <xorAxAx> grzywacz: "# FIXME: this depends on Jabber ATM, but may not do so in the future" should indeed move into a class attrib of the event handlers ...
2007-06-19T19:15:30  <xorAxAx> something like def can_handle(...): ...
2007-06-19T19:40:12  <lanius> ThomasWaldmann: http://www.cojobo.net/~h_wendel/stuff/exception.txt
2007-06-19T19:41:19  <xorAxAx> looks like thomas pushed incomplete and untested code from my student's repo to the main branch :)
2007-06-19T19:43:25  <xorAxAx> lanius: or did you pull from grzywacz?
2007-06-19T19:43:29  <ThomasWaldmann> ...to fix the problems caused by the code pushed by you :-P
2007-06-19T19:43:37  <xorAxAx> oh well
2007-06-19T19:44:01  <xorAxAx> so the only solution is to pull again :)
2007-06-19T19:44:14  <xorAxAx> lanius: pull from grzywacz and see if it goes away
2007-06-19T19:47:02  <lanius> i will just wait for you (anybody) to fix it ;)
2007-06-19T19:48:43  <xorAxAx> lanius: assume that it is fixed and try
2007-06-19T19:48:53  <xorAxAx> this is the idea of a distributed development model
2007-06-19T19:49:04  <xorAxAx> you can revert to your old state if it doesnt work
2007-06-19T19:49:11  <xorAxAx> think agile.
2007-06-19T19:49:12  <xorAxAx> :-)
2007-06-19T19:49:37  <lanius> that's what i did, now i wait for it been fixed in 1.7, then i try to merge again
2007-06-19T19:51:13  <lanius> since the last commit from grzywacz is 25 hours ago and the merge to 1.7 is 22 hour ago, i assume there is not yet a fix
2007-06-19T19:52:00  <xorAxAx> he will be here anyway :)
2007-06-19T20:22:18  * ThomasWaldmann tries to write some tests
2007-06-19T20:49:02  <grzywacz> What fix?
2007-06-19T20:49:32  <grzywacz> ooh
2007-06-19T20:51:56  <grzywacz> theuser.subscribed_events = [ev for ev in form.get('events', [])]
2007-06-19T20:52:02  <grzywacz> This should fix it, right?
2007-06-19T20:52:54  <xorAxAx> yes
2007-06-19T20:53:24  <grzywacz> Commiting & pushing.
2007-06-19T20:54:40  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2162:2a7b535b1e44 1.7-jabber-knowak/MoinMoin/_tests/test_xmlrpc_multicall.py: Add a test for xmlrpc Fault serialization.
2007-06-19T20:54:42  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2163:37e16cdc3a84 1.7-jabber-knowak/jabberbot/ (xmlrpcbot.py xmppbot.py): Make the output of GetPageInfo command human-readable
2007-06-19T20:54:43  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2164:f10336f74a6e 1.7-jabber-knowak/MoinMoin/events/emailnotify.py: Make function names pep8-compatible.
2007-06-19T20:54:44  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2165:8aad85519f76 1.7-jabber-knowak/MoinMoin/userform.py: Fix error occuring when user has no subscribed events.
2007-06-19T20:54:45  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2166:3cfc01315d3a 1.7-jabber-knowak/ (5 files in 4 dirs): Merge with my devel branch.
2007-06-19T20:57:45  <grzywacz> Sorry for this mistake. 8)
2007-06-19T20:57:49  <grzywacz> lanius, ping.
2007-06-19T20:58:15  <grzywacz> lanius, pull rev 8aad85519f7 from  ssh://hg@hg.thinkmo.de/moin/1.7-jabber-knowak
2007-06-19T22:05:15  <dreimark> that spammer on FeatureRequests/NewPageSubscription writes nonenxisting urls on the page
2007-06-19T22:05:46  <dreimark> and the words are no words
2007-06-19T22:06:15  <dreimark> neagulm: http://moinmoin.wikiwikiweb.de/FeatureRequests/NewPageSubscription?action=recall&rev=31 here you have spam
2007-06-19T22:06:52  <neagulm> dreimark, please see http://divmod.org/trac/wiki/DivmodReverend
2007-06-19T22:07:58  <neagulm> dreimark, I could create another hook for Reverend. This way a user would have a choice: a simple classifier like Reverend or a full-featured spam classifier
2007-06-19T22:08:39  <neagulm> dreimark, two classifiers would be good for testing
2007-06-19T22:14:52  <dreimark> neagulm: sounds interesting, let the user choose is one part. probably if its fast enough it could solve "detect the language of a document" always
2007-06-19T22:16:15  <dreimark> if we add spambayes for classifing spam then may be reverend could be used parallel for some special issues
2007-06-19T22:16:31  <neagulm> neagulm, yes. It would be good enough... For that I would consider also Orange (I proposed it at the very beginning of SoC). Orange is in some way a kind of Java Weka  for python
2007-06-19T22:17:06  <neagulm> Reverend uses similar tehniques to SB.
2007-06-19T22:18:37  <dreimark> neagulm: another idea is to train it to find pages written like who do you write like? e.g. Help pages
2007-06-19T22:18:55  <dreimark> then it could be used for search too
2007-06-19T22:19:20  <neagulm> we could a second instance of SB for that too
2007-06-19T22:19:31  <dreimark> ok
2007-06-19T22:19:37  <dreimark> what is with unicode
2007-06-19T22:19:52  <neagulm> with reverend ?
2007-06-19T22:20:00  <dreimark> yeah and orange
2007-06-19T22:21:08  <neagulm> Reverend (http://divmod.org:81/svn/Divmod/trunk/Reverend/reverend/thomas.py) seems to be ok
2007-06-19T22:24:31  <neagulm> with orange I'm not sure... I think not
2007-06-19T22:27:25  <neagulm> brb
2007-06-19T22:27:57  <dreimark> we need unicode and both need to go with python 2.3 too
2007-06-19T22:29:48  <dreimark> reverend requires Python 2.3 or later.
2007-06-19T22:34:53  <xorAxAx> grzywacz: you need to make the message localisable
2007-06-19T22:35:00  <xorAxAx> grzywacz: in xmppbot
2007-06-19T22:35:09  <xorAxAx> grzywacz: i.e. you have to generate it on the wikiside
2007-06-19T22:35:29  <xorAxAx> (same for any other user message)
2007-06-19T22:43:44  <neagulm> dreimark, A modified version of Reverend works even with older versions of python
2007-06-19T22:44:37  <dreimark> python2.3 is enough I believe current code does not go with 2.2
2007-06-19T22:44:42  <grzywacz> I know.
2007-06-19T22:44:57  <neagulm> dreimark, The best thing is that we could bundle it with MM. Just 2 files.
2007-06-19T22:45:55  <neagulm> dreimark, it could be the basic classifier used by default. A user that wants something better or different could easily switch the classifier in wikiconfig
2007-06-19T22:46:03  <dreimark> which license did reverend have
2007-06-19T22:47:34  <dreimark> ive seen it GNU Lesser General Public
2007-06-19T22:48:12  <neagulm> GNU Lesser General Public License
2007-06-19T22:49:03  <dreimark> ok, but spambayes needs to be tested too for working state
2007-06-19T22:50:56  <dreimark> the point "could easily switch the classifier in wikiconfig" needs to be tested
2007-06-19T22:52:26  <neagulm> dreimark, easily switch at install time not at runtime. The second option  would require deleting the cache, the db, etc
2007-06-19T22:52:37  <neagulm> but it's achivable
2007-06-19T23:07:50  <dreimark> neagulm: deleteing the cache is ok but why deleting the db, may be you could use the classifier name in the db name and avoid deleting if one like tro try out san other classifier
2007-06-19T23:08:16  <dreimark> he may be wants to go back to the one he likes and don't want to train again
2007-06-19T23:09:15  <neagulm> this would not be easy... Both classifiers use the same mathematical foundations but the way they select feature is different
2007-06-19T23:09:48  <neagulm> I will study the problem... I would be nice to be able to use the same db.
2007-06-19T23:11:56  <dreimark> neagulm: I was refering to "The second option  would require deleting the cache, the db,"
2007-06-19T23:13:00  <dreimark> same db would be nice, but why is it neccessary to delete it if you change the classifier?
2007-06-19T23:14:03  <neagulm> dreimark, it is not necessary as long SB and Reverend use the same db backend and have a similar db schema
2007-06-19T23:15:07  <dreimark> ahh, ok
2007-06-19T23:18:11  <dreimark> neagulm: it does not need to be named spam.db always  you can change it to revernd.db too
2007-06-19T23:18:34  <dreimark> so the classifier gives the name of the db
2007-06-19T23:18:59  <neagulm> yes
2007-06-19T23:19:01  <dreimark> and i would prefer the current db named spambayes.db
2007-06-19T23:19:30  <neagulm> ok
2007-06-19T23:24:31  <dreimark> neagulm: please check how often in the past they have changed theire db schema
2007-06-19T23:24:52  <dreimark> that is related to further updates because then we have to provide a mig script too
2007-06-19T23:25:48  <neagulm> ok. I will study the big changes to SB and Reverend (the storage part) in the last year
2007-06-19T23:30:00  <neagulm> dreimark, I'm going offline
2007-06-19T23:30:25  <dreimark> neagulm: ok, good night
2007-06-19T23:30:28  <neagulm> good night
2007-06-19T23:41:38  <zenhase> re
2007-06-19T23:43:28  <zenhase> sorry, i've been busy for the day
2007-06-19T23:44:48  <zenhase> regarding the design paper: i am still looking at the several parts where output is generated right now
2007-06-19T23:45:56  <zenhase> it's more complicated than i thought when i started this
2007-06-19T23:47:44  <zenhase> anyone of the senior developers still there?
2007-06-19T23:49:29  <ThomasWaldmann> btw, 3 weeks left until midterm eval
2007-06-19T23:49:41  <zenhase> i see
2007-06-19T23:51:05  <zenhase> i will do my best
2007-06-19T23:51:19  <xorAxAx> i am not a senio developer, i am a visionary conslutant
2007-06-19T23:52:15  <zenhase> i wasn't quite sure about the term to use
2007-06-19T23:52:38  <xorAxAx> how about "anybody of us in here?" :)