2007-04-21T00:02:29 <flowhase> re
2007-04-21T00:05:31 <dreimark> moin johill
2007-04-21T00:06:11 <johill> hi
2007-04-21T00:06:42 <johill> yeah, I'm having connectivity problems ;) my server is being moved to a different datacenter
2007-04-21T00:22:25 <dreimark> sometimes I have this trouble too
2007-04-21T00:23:03 <dreimark> switching systems and this is the reasone why I like to have more instantly created moin-dev logs
2007-04-21T00:45:47 <johill> grzywacz: "Communication from wiki process to the jabber bot will use a new, simple protocol. [...]"
2007-04-21T00:46:03 <johill> grzywacz: doesn't the jabber bot have to be trusted anyway to be allowed to do reverts etc?
2007-04-21T00:47:25 <grzywacz> johill, yes, it has to be trusted. It's just that unpickling wrong stuff leads directly to code execution
2007-04-21T00:48:24 <johill> ok so I guess you're saying code execution is so much worse than wiki damage
2007-04-21T00:49:43 <johill> you say xml is overkill, but wouldn't xmlrpc be easy to program with? as opposed to creating all the protocol handling yourself
2007-04-21T00:49:47 <xorAxAx> its a pity that python doesnt have a marshaller besides the xmlrpc one
2007-04-21T00:49:52 <grzywacz> If *both* can happen, I'd rather see it limited to wiki damage. ;)
2007-04-21T00:50:10 <xorAxAx> hmm, jabber is a bit of xml anywy
2007-04-21T00:50:24 <johill> a bit seems a bit of an understatement ;)
2007-04-21T00:50:27 <grzywacz> Sure.
2007-04-21T00:50:29 <grzywacz> ;]
2007-04-21T00:51:43 <johill> I guess I just wouldn't burden myself with that in this case
2007-04-21T00:52:22 <grzywacz> With what?
2007-04-21T00:52:34 <johill> all the marshalling for a new protocol
2007-04-21T00:52:51 <grzywacz> So you're suggesting to use xmlrpc? :)
2007-04-21T00:53:59 <xorAxAx> you could use the xmlrpc marshaller
2007-04-21T00:54:10 <xorAxAx> or depend on simplejson on both ends
2007-04-21T00:54:25 <xorAxAx> or marshall strings only
2007-04-21T00:54:50 <xorAxAx> in some simple packline/unpackline format :)
2007-04-21T00:55:00 <xorAxAx> (the syntax that the package installer is using :))
2007-04-21T00:55:15 <johill> then again, I guess xmlrpc can't do calls in both ways
2007-04-21T00:55:30 <xorAxAx> why not?
2007-04-21T00:55:42 <johill> can it? I don't really know
2007-04-21T00:55:49 <xorAxAx> grzywacz: indeed, it makes sense to use xmlrpc on the moin side
2007-04-21T00:56:06 <xorAxAx> grzywacz: otherwise cgi wouldnt be able to use it at all
2007-04-21T00:56:17 <johill> dang, using irssi keys in xchat doesn't work
2007-04-21T00:56:20 <xorAxAx> grzywacz: or think about fastcgi with 4 workr processes?
2007-04-21T00:56:32 <xorAxAx> grzywacz: which one of them would open which port?
2007-04-21T00:56:49 <johill> so what I was trying to say was that you probably can't have one open tcp xmlrpc connection and do many calls in both directions over it
2007-04-21T00:56:52 <grzywacz> Hm.
2007-04-21T00:57:01 <xorAxAx> of course you could select() the comm port in oyur additional thread
2007-04-21T00:57:11 <grzywacz> The way I've imagined this, would be moin->bot communication only
2007-04-21T00:57:30 <xorAxAx> grzywacz: how do you do any comm in the opposite direction then?
2007-04-21T00:57:32 <johill> but the bot has to say "please revert ..." etc
2007-04-21T00:57:36 <xorAxAx> like, new edit messages :)
2007-04-21T00:57:57 <johill> xorAxAx: select() and using the same socket from different threads seems quite daring
2007-04-21T00:57:59 <grzywacz> xorAxAx, it would use moin's code to perform edits? Kind of a jabber-moin-bot
2007-04-21T00:58:00 <grzywacz> ;]
2007-04-21T00:58:18 <xorAxAx> johill: why different threads?
2007-04-21T00:58:28 <johill> sounded like you were suggesting that
2007-04-21T00:58:35 <xorAxAx> grzywacz: you mean it should run on the same machine with file access?
2007-04-21T00:58:41 <xorAxAx> hmm
2007-04-21T00:58:49 <grzywacz> Yes, if it is to modify stuff.
2007-04-21T00:58:49 <johill> you were talking about having it on an untrusted network earlier ;)
2007-04-21T00:58:51 <johill> make up your mind
2007-04-21T00:58:57 <xorAxAx> hehe
2007-04-21T00:59:00 <grzywacz> bah
2007-04-21T00:59:06 <xorAxAx> johill: via nfs via smb of course!
2007-04-21T00:59:08 <xorAxAx> :-)
2007-04-21T00:59:11 <xorAxAx> and afs
2007-04-21T00:59:11 <grzywacz> notification-only bot can work on another machine
2007-04-21T00:59:22 <johill> xorAxAx: heh
2007-04-21T00:59:30 <grzywacz> heh indeed
2007-04-21T00:59:40 <grzywacz> I see no other way to do it right.
2007-04-21T00:59:50 <johill> edit locking will also be hard then
2007-04-21T00:59:50 <grzywacz> Oh, maybe writing a third moin-only bot would work, but that's just wrong.
2007-04-21T00:59:52 <xorAxAx> "you can deploy this bot anywhere you like, just for any real world usage with some funky features, it has to run on the same machine with the same user"
2007-04-21T01:00:03 <johill> I mean if somebody has an edit log but your jabber bot wants to revert something
2007-04-21T01:00:05 <xorAxAx> johill: thats a whole different issue
2007-04-21T01:00:06 <johill> that gets hairy
2007-04-21T01:00:12 <johill> s/log/lock/
2007-04-21T01:00:21 <grzywacz> Why? Jabber bot has to respect the lock.
2007-04-21T01:00:25 <johill> right
2007-04-21T01:00:33 <johill> but if it's on a different machine? does that work properly?
2007-04-21T01:00:41 <johill> eh I mean if it just accesses the storage
2007-04-21T01:00:43 <xorAxAx> moin doesnt depend on fnctl
2007-04-21T01:00:55 <grzywacz> I've already written that editing won't work if it's on a different machine.
2007-04-21T01:00:55 <xorAxAx> it has 3 different locking subsystems
2007-04-21T01:01:04 <johill> xorAxAx: yeah. bah
2007-04-21T01:01:11 <xorAxAx> isnt that enterprisey? :)
2007-04-21T01:01:14 <grzywacz> -_-;
2007-04-21T01:01:26 <johill> surely that's the only thing enterprisey means ;)
2007-04-21T01:01:35 <johill> or is that enterpricey?
2007-04-21T01:01:53 <grzywacz> It's not pricey at all. ;)
2007-04-21T01:02:14 <johill> hm
2007-04-21T01:02:22 <grzywacz> So, anyway. That's how I imagine this stuff:
2007-04-21T01:02:26 <johill> if the jabber process is long-running but moin is cgi?
2007-04-21T01:02:29 <grzywacz> notifications -> bot can run anywhere
2007-04-21T01:02:33 * johill is confused by all the options
2007-04-21T01:02:33 <grzywacz> editing -> local bot
2007-04-21T01:02:49 <johill> but the bot could use the existing xmlrpc interface to do editing :P
2007-04-21T01:02:52 <grzywacz> You can run two, why not. With different JIDs.
2007-04-21T01:02:55 <johill> -> bot anywhere
2007-04-21T01:03:00 <grzywacz> hm
2007-04-21T01:03:05 <grzywacz> Now, that's interesting.
2007-04-21T01:03:20 <johill> of course, then it has to log in as some user
2007-04-21T01:03:27 <grzywacz> Sure, but that makes sense.
2007-04-21T01:03:29 <johill> (once I fix the xmlrpc auth)
2007-04-21T01:03:54 <xorAxAx> yeah, it could use xmlrpc
2007-04-21T01:04:04 <johill> that's probably the best option anyway since then moin can be a cgi
2007-04-21T01:04:10 <grzywacz> This solution surely creates more overhead than a bot "understanding" moin.
2007-04-21T01:04:18 <grzywacz> But it's way cleaner as well.
2007-04-21T01:04:24 <xorAxAx> and you dont have user account issues
2007-04-21T01:04:27 <johill> but it's a lot more memory efficient too
2007-04-21T01:04:30 <xorAxAx> or machine-locking issues
2007-04-21T01:04:35 <xorAxAx> johill: hmm?
2007-04-21T01:04:41 <grzywacz> memory?
2007-04-21T01:04:44 <johill> not all the moin core code loaded twice
2007-04-21T01:04:50 <johill> (if you run it on the same machine)
2007-04-21T01:05:37 <xorAxAx> it isnt very large
2007-04-21T01:06:29 <johill> my standalone is about 8M
2007-04-21T01:07:00 <johill> 10 once you do some requests
2007-04-21T01:12:06 <grzywacz> Ook. I haven't though about xmlrpc before, but now that you've mentioned it, it sounds like a good idea.
2007-04-21T01:12:29 <grzywacz> It can work both ways as well. idea++;
2007-04-21T01:12:39 <grzywacz> Shame on me for not coming up with it. :)
2007-04-21T01:12:54 <xorAxAx> i am the mentor, i have to have the ideas :)
2007-04-21T01:13:03 <johill> not sure if you want to use it both ways
2007-04-21T01:13:12 <johill> notification is going to be much more common than editing
2007-04-21T01:13:28 <johill> so you want it to be much faster, and xmlrpc does have the new connection overhead in there
2007-04-21T01:13:35 <grzywacz> johill, that's true, but we may end up with less code this way ;)
2007-04-21T01:13:43 <johill> or can you keep xmlrpc connections open?
2007-04-21T01:13:52 <johill> nah, why would multicall exist then :)
2007-04-21T01:14:10 <grzywacz> yeah
2007-04-21T01:14:18 <johill> dunno, notification won't be something that takes a lot of data
2007-04-21T01:14:20 <xorAxAx> johill: thats an incorrect conclusion
2007-04-21T01:14:23 <johill> heh
2007-04-21T01:14:31 <xorAxAx> johill: multicall is one transaction
2007-04-21T01:14:38 <xorAxAx> HTTP pipelining/keep alive is unrelated
2007-04-21T01:14:44 <xorAxAx> thats one layer below
2007-04-21T01:14:49 <johill> true
2007-04-21T01:14:56 <xorAxAx> and depends on the quality of the client/server libs
2007-04-21T01:15:02 <johill> hm
2007-04-21T01:15:10 <johill> that would probably be fault-tolerant automatically
2007-04-21T01:15:25 <johill> i.e. if the other side is restarted you use a new connection, if the connection is still there use the old one
2007-04-21T01:15:30 <grzywacz> Yes.
2007-04-21T01:15:35 <johill> depends how it's implemented though
2007-04-21T01:16:45 <johill> I don't necessarily disagree with using xmlrpc both ways, in any case it's an easy prototype
2007-04-21T01:17:07 <xorAxAx> i think there are different bottlenecks than this xmlrpc one
2007-04-21T01:17:28 <xorAxAx> and if it turns out to be one, you can exchange the protocol if its nicely abstracted
2007-04-21T01:17:28 <johill> like what?
2007-04-21T01:17:33 <xorAxAx> (keep that in mind :))
2007-04-21T01:17:36 <johill> heh yeah
2007-04-21T01:17:39 <xorAxAx> johill: e.g. searching for subscribers
2007-04-21T01:17:44 <xorAxAx> it is O(#users) :-)
2007-04-21T01:17:48 <johill> well we have that already
2007-04-21T01:17:55 <xorAxAx> exactly
2007-04-21T01:18:08 <johill> ah ok, I see what you're saying
2007-04-21T01:18:45 <grzywacz> I'd like to start with having xmlrpc both ways, then. If it really becomes a problem, I can make a new, simple protocol for notifications only, no problems with that.
2007-04-21T01:18:54 <johill> yup
2007-04-21T01:19:08 <johill> btw
2007-04-21T01:19:15 <johill> we really should have a pylint config file
2007-04-21T01:19:37 <johill> it keeps complaining about abstract classes that only have one descendant
2007-04-21T01:19:44 <johill> or is there a better way to define an interface?
2007-04-21T01:20:39 <grzywacz> Contract? ;)
2007-04-21T01:21:13 <xorAxAx> johill: see Makefile
2007-04-21T01:21:29 <johill> ah
2007-04-21T01:21:34 <xorAxAx> johill: yes, using zope.interface <duck>
2007-04-21T01:21:42 <johill> heh
2007-04-21T01:21:57 <johill> haha
2007-04-21T01:22:00 <johill> make pylint
2007-04-21T01:22:01 <johill> ...
2007-04-21T01:22:04 <johill> pylint.utils.UnknownMessage: No such message id W0302
2007-04-21T01:22:25 <xorAxAx> version mismatch
2007-04-21T01:22:26 <johill> guess I need to update or something
2007-04-21T01:35:35 <grzywacz> But then, if I'm to write such a xmlrcp oriented bot, Twisted comes into play as an interesting option again. ;o
2007-04-21T01:35:58 <grzywacz> Or maybe I should not waste time with large frameworks and KISS? :D
2007-04-21T01:36:05 <johill> why twisted?
2007-04-21T01:36:13 <xorAxAx> python has a xmlrpc dispatcher out of the box
2007-04-21T01:36:26 <grzywacz> ACK
2007-04-21T01:36:32 <grzywacz> KISS, then.
2007-04-21T01:36:33 <xorAxAx> of course async stuff is better than threaded stuff
2007-04-21T01:36:43 <xorAxAx> but KISS is better than twisted
2007-04-21T01:37:05 <grzywacz> Very well. Let me put the result of this discussion on the wiki. :)
2007-04-21T01:37:13 <johill> some simpler async stuff would be nice
2007-04-21T01:37:37 <johill> maybe based on coroutines now that python2.5 supports those :)
2007-04-21T01:38:12 <johill> I once wrote a framework like that in pascal
2007-04-21T01:38:17 <xorAxAx> we are the knights of NIH NIH NIH!
2007-04-21T01:38:22 <grzywacz> gah ;]
2007-04-21T01:38:23 <xorAxAx> we want a twisted shrubbery!
2007-04-21T01:38:38 <grzywacz> johill, too bad I have to make it 2.3 compatible :D
2007-04-21T01:38:47 <xorAxAx> i mean, there are so many async IO schedulers for python
2007-04-21T01:38:48 <johill> heh
2007-04-21T01:38:53 <xorAxAx> i wrote one myself, its in moin
2007-04-21T01:38:56 <johill> are there?
2007-04-21T01:39:05 <xorAxAx> its used to batch the wikisync calls
2007-04-21T01:39:13 <xorAxAx> johill: i assume so
2007-04-21T01:39:29 <johill> async io scheduler to batch wikisync? hmm
2007-04-21T01:41:01 <johill> heh
2007-04-21T01:41:04 <johill> here's an idea
2007-04-21T01:41:08 <johill> instead of doing
2007-04-21T01:41:14 <johill> read(x) and write(x)
2007-04-21T01:41:37 <johill> do yield read,x or yield write,x and then ...
2007-04-21T01:41:45 <xorAxAx> johill: thats how it works
2007-04-21T01:41:52 <xorAxAx> you are a year too late
2007-04-21T01:41:53 <johill> of course that doesn't really work well since you can't actually call functions
2007-04-21T01:41:54 <xorAxAx> :-)
2007-04-21T01:42:00 <johill> that do some io
2007-04-21T01:42:14 <johill> :)
2007-04-21T01:42:33 <johill> my earlier stuff was based on fibers
2007-04-21T01:42:36 <xorAxAx> it is doing yield server_proxy.prepare_call_read(my parameters here)
2007-04-21T01:42:44 <johill> is that that the windows term?
2007-04-21T01:42:49 <johill> the unix term is something else I think
2007-04-21T01:42:51 <xorAxAx> windows calls them like that
2007-04-21T01:42:56 <xorAxAx> user space threads
2007-04-21T01:43:00 <johill> no
2007-04-21T01:43:14 <xorAxAx> well, they are nothing else :)
2007-04-21T01:43:23 <johill> yeah, but they have a different name
2007-04-21T01:44:22 <johill> anyway, it works really well, but I guess python can't really do that natively
2007-04-21T01:44:37 <xorAxAx> johill: cf. greenlets
2007-04-21T01:44:40 <xorAxAx> work very nicely
2007-04-21T01:44:50 <xorAxAx> __doc__ wrote a wsgi compliant server based on them
2007-04-21T01:44:54 <xorAxAx> 2000 reqs/s
2007-04-21T01:45:07 <xorAxAx> greenlets were crafted by armin rigo
2007-04-21T01:45:20 <johill> nice, how is it implemented?
2007-04-21T01:45:23 <xorAxAx> who took code from tismer's stackless python to do the stackframe switching
2007-04-21T01:45:37 <xorAxAx> in C :)
2007-04-21T01:45:48 <johill> well in C doesn't say much about how ;)
2007-04-21T01:45:50 <xorAxAx> it is copying the C stack away
2007-04-21T01:46:12 <grzywacz> xorAxAx, anyway, I've decided to use pyxmpp. It has the largest feature set and is almost fully xmpp compliant, there's been some work done on it not-so-long-ago, it works with python 2.3
2007-04-21T01:46:28 <grzywacz> Twisted might work, but it lacks certain stuff and has no release schedule available.
2007-04-21T01:46:34 <xorAxAx> grzywacz: nice to know. and if there is no work, you might be able to ask for commit priveledges :)
2007-04-21T01:47:06 <grzywacz> Actually, pyxmpp's author wrote a comment on my blog a few days ago ;-)
2007-04-21T01:47:43 <xorAxAx> hehe
2007-04-21T01:49:31 <johill> I suppose my multistage auth could be implemented a lot simpler in stackless python ;)
2007-04-21T01:49:35 <grzywacz> Anyway, how's work going on the templating front?
2007-04-21T01:49:45 <xorAxAx> lets ask flowhase
2007-04-21T01:49:47 <johill> just pickle the tasklet and bingo
2007-04-21T01:50:16 <xorAxAx> johill: did you plan to write any unittests btw? :)
2007-04-21T01:50:27 <xorAxAx> makes it much more probable to survive future commits
2007-04-21T01:50:44 <johill> yeah, I did indeed plan to, but need to look into that py.test stuff you did
2007-04-21T01:51:17 <xorAxAx> well, the py.test docs are quite nice, then you simply create a file that has a class called TestFoo
2007-04-21T01:51:33 <xorAxAx> with a method that can use self.request like before
2007-04-21T01:51:38 <xorAxAx> and then you can use assert
2007-04-21T01:51:48 <xorAxAx> or import py; py.test.raises
2007-04-21T01:51:54 <johill> yeah, I was reading those before I went dancing
2007-04-21T01:51:59 <xorAxAx> hehe
2007-04-21T01:52:35 <grzywacz> I can't get a tick icon to show up with "(./)" :\
2007-04-21T01:53:28 <grzywacz> Ah, it doesn't work inside headers. Oh well.
2007-04-21T01:53:58 <johill> nothing works inside headers
2007-04-21T01:56:16 <grzywacz> catarrh is killing me :\
2007-04-21T01:58:28 <johill> reason: Skipped: Many broken tests, much broken code, broken, broken, broken.
2007-04-21T01:58:29 <johill> heh
2007-04-21T01:59:04 <xorAxAx> :-)
2007-04-21T02:00:36 <grzywacz> Wiki updated.
2007-04-21T02:00:43 * grzywacz dives into python's xmlrpc docs
2007-04-21T02:01:42 <xorAxAx> good that we didnt accept the other polish student in our soc queue ( the one that i sold to pypy)
2007-04-21T02:01:52 <grzywacz> xorAxAx, why good?
2007-04-21T02:01:55 <xorAxAx> he hasnt shown up a single time in pypy irc yet after he was accepted :)
2007-04-21T02:02:01 <grzywacz> oww
2007-04-21T02:02:24 <grzywacz> I'm used to working in open source environment. ;)
2007-04-21T02:02:29 <xorAxAx> hehe :-)
2007-04-21T02:02:55 <grzywacz> Do you happen to live anywhere near Dusseldorf? ;)
2007-04-21T02:03:33 <johill> depends on your scale ;)
2007-04-21T02:03:39 <grzywacz> heh ;)
2007-04-21T02:03:57 <grzywacz> Near enough to make travel reasonably possible. :D
2007-04-21T02:04:47 <xorAxAx> duesseldorf is 2 h by train from here
2007-04-21T02:04:52 <johill> about 2-3 hours by train
2007-04-21T02:04:55 <grzywacz> Hm, not bad.
2007-04-21T02:05:08 <xorAxAx> but only because the (cheap) train is fast
2007-04-21T02:05:22 <xorAxAx> the last time i was in duesseldorf was a pypy sprint :)
2007-04-21T02:05:28 <grzywacz> Asking out of curiosity because we're making a Wesnoth conference at the and of August in Dusseldorf. :P
2007-04-21T02:05:28 <xorAxAx> for one day ...
2007-04-21T02:05:34 <xorAxAx> cool
2007-04-21T02:05:34 <grzywacz> s/and/end/
2007-04-21T02:05:56 <xorAxAx> there are times when i was patient
2007-04-21T02:05:57 <johill> heh, the last time I was there must have been for some beers with ex-colleagues of my g/f ;)
2007-04-21T02:06:02 <xorAxAx> playing wesnoth for hours
2007-04-21T02:06:07 <xorAxAx> s/are/were/
2007-04-21T02:06:42 * johill looks at the xmlrpc auth code again
2007-04-21T02:06:55 <grzywacz> xorAxAx, when was that?
2007-04-21T02:07:11 <grzywacz> The game has been improving steadily. :)
2007-04-21T02:07:28 <xorAxAx> grzywacz: hmm, half a year ago?
2007-04-21T02:07:36 <grzywacz> Ah, ok. :)
2007-04-21T02:07:38 <xorAxAx> do you mean the game play?
2007-04-21T02:07:40 <xorAxAx> it got faster?
2007-04-21T02:07:47 <xorAxAx> i cant imagine how to achieve that
2007-04-21T02:07:51 <grzywacz> Gameplay? No.
2007-04-21T02:07:56 <grzywacz> But the interface is smoother now.
2007-04-21T02:08:01 <xorAxAx> well, if you have people who need 5 minutes to complete a move
2007-04-21T02:08:18 <grzywacz> And development done in the dev branch is quite interesting.
2007-04-21T02:08:19 <xorAxAx> also, the help window needed 8 seconds to load or something like that
2007-04-21T02:08:28 <xorAxAx> ah, what will be in the next version? :)
2007-04-21T02:08:34 <grzywacz> What's your hardware?
2007-04-21T02:08:49 <xorAxAx> hmm, 1,6 GHz, geforce 4 on that machine
2007-04-21T02:08:52 <grzywacz> xorAxAx, well, wesnoth has a release early & release often policy ;)
2007-04-21T02:08:58 <grzywacz> Weird, then.
2007-04-21T02:09:07 <xorAxAx> windows
2007-04-21T02:09:17 <xorAxAx> hmm, let me try on this linux machine
2007-04-21T02:09:23 <grzywacz> But yeah, scrolling has been vastly improved and tweaked font cache may reduce lag in help and other dialogs as well.
2007-04-21T02:10:04 <grzywacz> Units are getting more and more animations... Idle ones look fancy.
2007-04-21T02:10:31 <grzywacz> That reminds me that I have to finish my sound sources code. :\
2007-04-21T02:10:31 <xorAxAx> cool
2007-04-21T02:11:27 <xorAxAx> ah, and i remembr that the chat code was O(n^2)
2007-04-21T02:11:30 <xorAxAx> for something
2007-04-21T02:11:33 <xorAxAx> nick list updating? :)
2007-04-21T02:11:46 <grzywacz> Maybe.
2007-04-21T02:11:51 <grzywacz> Game list updating?
2007-04-21T02:12:19 <grzywacz> Or maybe it still is, but optimisations have removed any noticeable lag.
2007-04-21T02:12:49 <grzywacz> Who cares about O(n^2) when you update a list of 40 games every 5 seconds. :P
2007-04-21T02:13:43 <johill> why are there like three or four ssl kits for python?
2007-04-21T02:14:07 <xorAxAx> grzywacz: it was really sluggish
2007-04-21T02:14:11 <xorAxAx> and annoying
2007-04-21T02:14:16 <grzywacz> xorAxAx, I know, it's been fixed.
2007-04-21T02:15:20 <grzywacz> |Kev|, reminding you as promised: http://moinmoin.wikiwikiweb.de/JabberSupport
2007-04-21T02:17:12 * johill thinks about a jabber based openid service
2007-04-21T02:17:41 <grzywacz> johill, http://wiki.jabber.org/index.php/HTTP-Auth_suite
2007-04-21T02:19:06 <johill> hm, cool, combining that with a moin wiki that does http auth and you're 80% there
2007-04-21T02:20:06 <johill> I'd have done that differently I think and tried getting away without a firefox extension
2007-04-21T02:20:07 <johill> oh well
2007-04-21T02:20:13 <johill> once your code works I'll try that for moin wikis :)
2007-04-21T02:21:14 <grzywacz> does xmlrpc auth you've mentioned use http basic auth or some custom routines?
2007-04-21T02:21:29 <johill> both, iirc
2007-04-21T02:21:36 <xorAxAx> xmlrpc with http basic auth is not standardised
2007-04-21T02:21:41 <johill> custom works in any case (get token/use token)
2007-04-21T02:22:36 <johill> grzywacz: keep in mind that I intend to rewrite the xmlrpc auth token stuff to have tokens that expire and can be revoked
2007-04-21T02:22:50 <grzywacz> johill, any timeline?
2007-04-21T02:23:09 <johill> not before I sleep
2007-04-21T02:23:24 <grzywacz> No problem, then. 8)
2007-04-21T02:23:49 <johill> just treat the auth token as transitional
2007-04-21T02:23:54 <grzywacz> sure
2007-04-21T02:23:59 <johill> like that it might expire in half an hour or whatever
2007-04-21T02:31:28 <johill> MoinMoin:JohannesBerg/JabberAuthentication
2007-04-21T02:31:41 <johill> just a braindump of what I was actually thinking
2007-04-21T02:34:03 <grzywacz> Looks ok.
2007-04-21T02:36:20 <johill> it's been done, but the site (http://openid.xmpp.za.net/) is down
2007-04-21T02:41:02 <johill> _keturn: jyte.com doesn't like my openid server either
2007-04-21T02:42:34 <johill> same with botbouncer
2007-04-21T02:42:34 <johill> hrm
2007-04-21T02:44:29 <grzywacz> :>
2007-04-21T02:46:24 <johill> ehm, I need sleep
2007-04-21T02:46:31 <johill> might be because my server and thus my openid is down ;)
2007-04-21T02:46:40 <johill> but botbouncer didn't like me earlier when it was up, so...
2007-04-21T02:46:52 <grzywacz> good night johill :)
2007-04-21T02:48:35 <grzywacz> I think I should get some sleep as well... Good night.
2007-04-21T02:50:00 <xorAxAx> good night
2007-04-21T02:50:33 <xorAxAx> duesseldorf is 8 EUR btw with a bahncard 25
2007-04-21T02:50:42 <xorAxAx> johill: are you still a student? :)
2007-04-21T02:50:55 <johill> I have a bahncard 50 ;)
2007-04-21T02:51:00 <johill> yeah, I am though
2007-04-21T02:51:07 <xorAxAx> wow, then its even cheaper
2007-04-21T02:51:17 <xorAxAx> well, hamm is most of the way
2007-04-21T02:51:29 <xorAxAx> hmm, not fully, but its a lot at least
2007-04-21T02:52:29 <xorAxAx> ah, nearly at the middle
2007-04-21T02:54:25 <johill> I know how to get there, my g/f was there for a year ;)
2007-04-21T02:56:51 <johill> ok, the last todo item was getting the CSS right for openid to display the icon :)
2007-04-21T02:57:03 <johill> (./)
2007-04-21T03:37:12 <CIA-28> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2018:dbf06dea00aa 1.7/MoinMoin/auth/__init__.py: MoinLogin auth: continue if a previous auth was successful
2007-04-21T04:45:49 <flowhase> oh boy
2007-04-21T04:46:09 * flowhase catches up on the backlog
2007-04-21T04:50:17 <flowhase> grzywacz: i already thought that you would be interested in templating and therefore output generation too :>
2007-04-21T04:51:47 <flowhase> i am going to visit a friend in freiburg tomorrow for a birthday party but i'll be back sunday
2007-04-21T04:53:31 <flowhase> i hope fabi is back then too, i would start brainstorming then
2007-04-21T04:54:01 <flowhase> (until now it's just many loose ideas)
2007-04-21T04:55:50 <flowhase> well, need to get some sleep now, see you tomorrow noon ... or else sunday
2007-04-21T04:57:53 <flowhase> grzywacz: hmm, i like the style of your project wiki page
2007-04-21T04:59:03 <flowhase> i think i will adopt it a little and structure my ideas and progress tracking accordingly
2007-04-21T04:59:14 <flowhase> well then guys, good night
2007-04-21T10:40:44 <Fabi> aloha
2007-04-21T11:08:35 <xorAxAx> hi Fabi
2007-04-21T11:20:32 * johill installs the new auth stuff on linuxwireless.org et al. eat-your-own-dogfood style ;)
2007-04-21T11:24:21 <grzywacz> moin
2007-04-21T11:27:08 <xorAxAx> moin grzywacz
2007-04-21T11:33:43 <grzywacz> hi xorAxAx
2007-04-21T11:43:51 <johill> where does moin logging go in fcgi?
2007-04-21T11:45:14 <xorAxAx> apache error log
2007-04-21T11:47:53 <johill> weird, don't see anything there
2007-04-21T11:50:20 <johill> http://moin-test.sipsolutions.net/ -> openid enabled moin wiki
2007-04-21T11:51:11 <xorAxAx> and the bot still dislikes you?
2007-04-21T11:51:23 <johill> bot?
2007-04-21T11:51:27 <johill> go to login on that page :)
2007-04-21T11:51:35 <xorAxAx> the testing bot site
2007-04-21T11:51:48 <johill> oh yeah
2007-04-21T11:51:54 <johill> botbouncer you mean
2007-04-21T11:52:07 <johill> it's not a bot, it's a capcha thingie
2007-04-21T11:52:13 <johill> anyway, lunch
2007-04-21T12:15:15 <johill> hm
2007-04-21T12:15:26 <johill> it's a bit stupid to have 'create a user now' when the wiki only supports openid login ;)
2007-04-21T12:18:10 <ThomasWaldmann> johill: there is no .trail in 1.6 (nor 1.7)
2007-04-21T12:18:36 <johill> hm?
2007-04-21T12:18:53 <johill> -EMISSINGCONTEXT
2007-04-21T12:19:42 <johill> changeset 2010?
2007-04-21T12:21:50 <ThomasWaldmann> your 1.7 help page
2007-04-21T12:23:27 <johill> ah ok, will take a look
2007-04-21T12:29:51 <johill> ThomasWaldmann: ok, removed mention of .trail on both moinmaster and the todo page
2007-04-21T12:35:32 <ThomasWaldmann> thanks
2007-04-21T12:41:12 <johill> ThomasWaldmann: if you haven't seen, I put up my openid stuff on moin-test.sipsolutions.net, feel free to experiment there
2007-04-21T12:41:32 <johill> I might enable it for linuxwireless.org too
2007-04-21T12:50:43 <johill> harr
2007-04-21T12:50:48 <johill> botbouncer works now
2007-04-21T12:50:53 <johill> it still uses the wrong spec ID URL
2007-04-21T12:51:02 <johill> well, I can advertise both :)
2007-04-21T12:51:48 <ThomasWaldmann> will try do do that later, gtg soon
2007-04-21T12:57:52 <johill> and jyte.com works too
2007-04-21T12:57:56 * johill slaps _keturn
2007-04-21T13:05:29 <johill> uh oh
2007-04-21T13:05:35 <johill> https://verify.sxip.com/email/ gets a nullpointer exception on their side
2007-04-21T13:05:35 <dreimark> johill: so we have a chance to get the OpenID Code Bounties
2007-04-21T13:05:44 <johill> https://verify.sxip.com/demorp/ gets a problem on this side
2007-04-21T13:05:56 <johill> I already asked about that but it seems pretty dead
2007-04-21T13:06:00 <johill> but yeah
2007-04-21T13:07:13 <johill> both of them must be problems with pyopenid
2007-04-21T13:11:34 <ThomasWaldmann> maybe they had a mutiny on their bounty :-P
2007-04-21T13:13:46 <dreimark> hehe
2007-04-21T13:57:01 <johill> hm
2007-04-21T13:57:26 <johill> I wonder if I broke auth_attribs
2007-04-21T13:57:47 <johill> probably
2007-04-21T13:58:54 <johill> yeah, dang
2007-04-21T14:03:39 <CIA-28> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2019:149573c7ecaf 1.7/MoinMoin/session.py: session must store user.auth_attribs
2007-04-21T14:07:04 <johill> what's the point of cfg.user_autocreate?
2007-04-21T14:07:12 <johill> if you have ldap configured you always want that, no?
2007-04-21T14:07:39 <johill> or it should be an option to ldap (only already known users)
2007-04-21T14:25:09 <johill> ugh, auth docs are really strange
2007-04-21T14:28:18 * johill rewrites it
2007-04-21T14:34:45 <johill> and HelpOnSession too
2007-04-21T14:36:38 <johill> for 1.7 anyway
2007-04-21T14:39:20 <johill> hmmm
2007-04-21T14:39:21 <johill> here's a thought
2007-04-21T14:39:27 <johill> extend ACLs to allow checking auth_method
2007-04-21T14:39:40 <johill> so say users that log in with username/password can read, but need ssl client cert to write
2007-04-21T14:42:41 <xorAxAx> johill: the system already supports that
2007-04-21T14:42:53 <xorAxAx> you would need to be able to tag auth methods as trusted/untrusted
2007-04-21T14:43:07 <xorAxAx> and modify the trusted check to look at those tags
2007-04-21T14:43:38 <johill> ah I just saw that too
2007-04-21T14:43:49 <johill> that would be easy to do, I'd think
2007-04-21T14:44:17 <johill> but it only supports trusted/not-trusted, not JohannesBerg(trusted) etc
2007-04-21T14:44:21 <johill> good enough I guess
2007-04-21T14:45:12 <johill> yeah so the idea would be to mark auth methods as trusted
2007-04-21T14:45:14 <johill> how does trusted work?
2007-04-21T14:45:19 * johill reads code
2007-04-21T14:45:46 <johill> hmmm
2007-04-21T14:45:48 <johill> are you sure?
2007-04-21T14:45:50 <johill> user.py says:
2007-04-21T14:45:58 <johill> valid, changed = self._validatePassword(user_data)
2007-04-21T14:46:00 <johill> if valid: trusted=1
2007-04-21T14:46:02 <johill> basically
2007-04-21T14:46:39 <xorAxAx> umm, then it was changed
2007-04-21T14:46:44 <xorAxAx> in 1.3, trusted meant "http auth"
2007-04-21T14:46:48 * xorAxAx sighs
2007-04-21T14:47:04 <johill> let's rip it out then and replace it with trusted_auth_methods
2007-04-21T14:48:43 <johill> you might want to fix HelpOnAccessControlLists
2007-04-21T14:48:46 <johill> or the code in 1.6
2007-04-21T14:52:08 <johill> it says
2007-04-21T14:52:14 <johill> def _special_Trusted(self, request, name, dowhat, rightsdict):
2007-04-21T14:52:14 <johill> """ check if user <name> is known AND even has logged in using a password.
2007-04-21T14:52:17 <johill> does not work for subsription emails that should be sent to <user>,
2007-04-21T14:52:21 <johill> as he is not logged in in that case.
2007-04-21T14:52:23 <johill> """
2007-04-21T14:52:26 <johill> that seems useless
2007-04-21T14:52:27 <johill> since you pretty much always need a password
2007-04-21T14:52:37 <johill> (unless you use a client cert but hey that's even more secure)
2007-04-21T14:54:34 <johill> but now where's the difference between known and trusted?
2007-04-21T14:56:31 <xorAxAx> hmm?
2007-04-21T14:56:43 <xorAxAx> you can define Trusted:read,write Known:red
2007-04-21T14:56:45 <xorAxAx> read
2007-04-21T14:56:58 <johill> yeah but in what case is a user actually known but not trusted?
2007-04-21T14:57:44 <xorAxAx> if the auth service is not trusted
2007-04-21T14:58:13 <johill> that's the theory
2007-04-21T14:58:25 <xorAxAx> hmm?
2007-04-21T14:58:33 <xorAxAx> in 1.3, it was like i said
2007-04-21T14:58:42 <xorAxAx> no idea how much sense there is currently
2007-04-21T14:58:43 <johill> the code right now in 1.6 sets trusted if user.name == user.auth_name
2007-04-21T14:58:46 <johill> (don't understand)
2007-04-21T14:58:47 <xorAxAx> maybe ThomasWaldmann knows
2007-04-21T14:58:54 <johill> and if self._validatePassword returned true
2007-04-21T14:58:57 <johill> (don't understand either)
2007-04-21T14:59:02 <johill> and that's it
2007-04-21T14:59:07 <xorAxAx> security by obscurity :-)
2007-04-21T14:59:07 <johill> no auth method involvement
2007-04-21T14:59:23 <johill> so how about JohannesBerg/AuthMethodACL
2007-04-21T15:00:09 <xorAxAx> johill: how would you identify the methods in cfg.trusted_auth_methods?
2007-04-21T15:00:26 <johill> by their name
2007-04-21T15:00:30 <johill> each auth method has a name now
2007-04-21T15:00:42 <xorAxAx> instance name?
2007-04-21T15:00:44 <xorAxAx> or method name?
2007-04-21T15:00:56 <xorAxAx> the former is important, you may need to auth method handlers
2007-04-21T15:00:57 <johill> neither, each auth method instance is required to have a .name value
2007-04-21T15:01:03 <xorAxAx> ah
2007-04-21T15:01:10 <xorAxAx> that is an instance name :)
2007-04-21T15:01:17 <johill> yeah I guess
2007-04-21T15:01:25 <xorAxAx> s/to au/to have multiple au/
2007-04-21T15:01:36 <johill> actually, it's more of a class name
2007-04-21T15:01:44 <xorAxAx> thats useless :)
2007-04-21T15:01:46 <johill> since it's usually not per-instance since they are singletons anyway
2007-04-21T15:01:52 <johill> no
2007-04-21T15:01:58 <xorAxAx> but they shouldnt be, imagine 2 ldap servers
2007-04-21T15:02:09 <johill> oh yeah you can do that in theory
2007-04-21T15:02:14 <xorAxAx> or two user directories of moin
2007-04-21T15:02:15 <johill> just set .name of the second one to something else
2007-04-21T15:02:20 <johill> like 'ldap2'
2007-04-21T15:02:24 <johill> that's fine
2007-04-21T15:02:41 <johill> but the default is that it's a class variable just so it's easier to code (remember that instance vars hide class vars)
2007-04-21T15:03:16 <dreimark> Is the difference between trusted and known: trusted logged in by password, known used the cookie
2007-04-21T15:03:24 <johill> heh
2007-04-21T15:03:30 <johill> how is that useful?
2007-04-21T15:04:03 <dreimark> no idea
2007-04-21T15:04:19 <johill> also, that distinction has just disappeared with my session/auth refactoring
2007-04-21T15:04:37 <johill> no actually, it hasn't, but trusted will be set only once
2007-04-21T15:04:50 <johill> unless you used http auth and resend the password every time
2007-04-21T15:04:52 <johill> haha
2007-04-21T15:05:05 <johill> so there we have the same semantics in a very complicated way
2007-04-21T15:05:24 <xorAxAx> 1.3 semantics, yes
2007-04-21T15:05:26 <johill> ok let's remove that and maybe make trusted_auth_methods default to ['http']
2007-04-21T15:05:55 <johill> (not that I trust http auth.. resending the password all the time...)
2007-04-21T15:06:56 <johill> alright, I'll make a patch
2007-04-21T15:15:22 <johill> xorAxAx: does xmlrpc need trusted?
2007-04-21T15:15:26 <johill> it only seems to use it in whoami
2007-04-21T15:15:48 <johill> oh no, it does use it elsewhere, n/m
2007-04-21T15:21:09 <CIA-28> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2020:1b0629547090 1.7/MoinMoin/ (6 files in 4 dirs): introduce cfg.trusted_auth_methods
2007-04-21T15:22:19 <CIA-28> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2021:927e97ed7d7c 1.7/MoinMoin/config/multiconfig.py: fix wrong default for trusted_auth_methods
2007-04-21T15:23:32 <johill> wouldn't it be sufficient for xmlrpc to use something akin to Known: instead of Trusted:?
2007-04-21T15:42:36 <johill> oh humm
2007-04-21T15:42:53 <johill> I think my openid stuff is incompatible with url_prefix_action
2007-04-21T15:44:55 <johill> but then again, I guess that even without /action/ in there things are still accepted so I should be fine, right?
2007-04-21T16:37:35 <CIA-28> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2022:c9a295039bf7 1.7/docs/CHANGES: update CHANGES
2007-04-21T17:18:56 * johill rediscovers PluggableACLs
2007-04-21T17:24:18 <johill> typical nir-idea-bashing in there... bah. makes me sick
2007-04-21T17:27:06 <xorAxAx> :-)
2007-04-21T17:27:18 <xorAxAx> the old times
2007-04-21T17:29:44 <johill> :)
2007-04-21T18:05:43 * johill ponders adding a botbouncer auth plugin
2007-04-21T18:12:39 <johill> should be easy
2007-04-21T18:43:22 <johill> works
2007-04-21T18:48:32 <ThomasWaldmann> moin
2007-04-21T18:48:54 <johill> moin
2007-04-21T18:50:43 <johill> cool, it really works
2007-04-21T18:51:37 <keturn> oh, it is true that jyte and botbouncer do look for yadis, but don't look for 2.0 anything
2007-04-21T18:51:51 <keturn> so if your yadis only advertises 2.0, they will not find anything there
2007-04-21T18:52:11 <johill> keturn: well they do work with the old 2.0 url
2007-04-21T18:52:29 <johill> should I have 1.1 in there too?
2007-04-21T18:52:35 <johill> but my botbouncer plugin works now ;)
2007-04-21T18:52:39 * johill wohoos
2007-04-21T18:53:12 <johill> now you can have captchas in moin by using botbouncer :)
2007-04-21T18:54:08 <johill> http://johannes.sipsolutions.net/patches/moin/all/2007-04-21-16:53/botbouncer.patch
2007-04-21T18:54:47 * johill ponders adding a 'please redirect the user to X' to the auth framework
2007-04-21T18:54:55 <johill> so I can get rid of the XXX request internals at the very bottom
2007-04-21T18:55:08 <johill> but the return value for login() is already a tuple with 4 items :(
2007-04-21T18:55:34 <xorAxAx> how about introducing a state object?
2007-04-21T18:55:54 <johill> keturn: wish item for botbouncer: let us control the green intro text up there or at least add some text
2007-04-21T18:55:58 <johill> xorAxAx: explain
2007-04-21T18:56:14 <xorAxAx> well, i mean some class that you return instead of 10-item tuples
2007-04-21T18:56:25 <johill> oh
2007-04-21T18:56:30 <johill> yeah, that's probably a good idea
2007-04-21T18:57:59 <keturn> they work with the old 2.0 url? I wouldn't have expected that. I'll have to talk to brian and find out what code those are running these days.
2007-04-21T18:58:57 <johill> keturn: I put in http://openid.net/signon/2.0 and they both work now
2007-04-21T18:59:19 <xorAxAx> can moin serve openid identities, johill?
2007-04-21T18:59:22 <johill> but should I expect to have yadis users for 1.0/1.1 and advertise those types?
2007-04-21T18:59:29 <johill> xorAxAx: yes, I have a patch for both ways
2007-04-21T18:59:38 <johill> serve and use
2007-04-21T18:59:52 <johill> I had serve before use even ;)
2007-04-21T19:02:41 <johill> xorAxAx: see MoinMoin:JohannesBerg/OpenID_support
2007-04-21T19:02:50 <johill> and http://moin-test.sipsolutions.net/
2007-04-21T19:02:54 <keturn> johill: yeah, probably
2007-04-21T19:03:14 <johill> ok, I'll add those then
2007-04-21T19:03:21 <keturn> in fact, I'd expect 1.0/1.1 to be more useful than /signon/2.0
2007-04-21T19:04:56 <johill> but /signon/2.0 made botbouncer work ;)
2007-04-21T19:05:06 <johill> what's thedifference between OPENID_IDP_2_0_TYPE and OPENID_2_0_TYPE?
2007-04-21T19:05:22 <xorAxAx> i suspect that IDP is the difference
2007-04-21T19:05:28 <xorAxAx> :-)
2007-04-21T19:05:54 <johill> back in a bit
2007-04-21T19:07:00 <keturn> xorAxAx figured it out ;)
2007-04-21T19:07:26 <xorAxAx> but i am not completely sure
2007-04-21T19:08:21 <ThomasWaldmann> btw, are auth modules already plugins? :)
2007-04-21T19:11:33 <keturn> e.g. http://myopenid.com/ will advertise itself as the IDP type, whereas http://keturn.myopenid.com/ doesn't, as the latter is a personal identifier, not the server's identifier.
2007-04-21T19:11:40 <Fabi> bbl
2007-04-21T19:45:29 <johill> ThomasWaldmann: uh, no, they aren't
2007-04-21T19:45:42 <johill> keturn: ah ok, so I don't care about the IDP really, do I?
2007-04-21T19:46:33 <ThomasWaldmann> johill: i just thought they might be, auth things seem to develop fast these days :)
2007-04-21T19:46:53 <johill> heh
2007-04-21T19:46:57 <johill> yeah why not
2007-04-21T19:47:01 <ThomasWaldmann> but maybe we need a farm plugin dir first
2007-04-21T19:47:04 <johill> then again
2007-04-21T19:47:06 <johill> no, why bother
2007-04-21T19:47:10 <johill> auth is defined in the config
2007-04-21T19:47:17 <johill> so you can load them from anywhere anyway
2007-04-21T19:47:39 <johill> you don't need to import from MoinMoin.auth you can just as well import from just openidrp.py wherever that may be
2007-04-21T19:48:00 <ThomasWaldmann> sure, just thought about making this import stuff easier
2007-04-21T19:48:22 <ThomasWaldmann> but it's not that important
2007-04-21T19:48:37 <johill> it wouldn't really be easier
2007-04-21T19:49:05 <johill> since plugins are loaded by name from some moin code, but auth stuff must be laoded by the admin in the config
2007-04-21T19:49:37 <ThomasWaldmann> like http = AuthPlugin('http', ...)
2007-04-21T19:49:40 <johill> keturn: what's the advantage of advertising the IDP type? why do RPs bother?
2007-04-21T19:49:47 <ThomasWaldmann> auth = [http, ...]
2007-04-21T19:49:50 <johill> ThomasWaldmann: ohh, ok, yeah that'd work
2007-04-21T19:50:07 <keturn> so you can type in your idp url instead of your whole identifier
2007-04-21T19:50:10 <johill> but I don't see much point in that
2007-04-21T19:50:32 <johill> keturn: ah. what do I have to do to support that?
2007-04-21T19:50:47 <johill> i.e. what kind of request would I get? some sort of user-choice request?
2007-04-21T19:51:04 <johill> (with moin I require previous login currently so I already know the user from the session)
2007-04-21T19:51:14 <keturn> yeah, you'd get a request with a special identifier_select URI
2007-04-21T19:51:27 <johill> let me check the specs if that's easy to do
2007-04-21T19:52:18 <johill> how does the lib represent that?
2007-04-21T19:52:44 <johill> ah that's just given you as the identity?
2007-04-21T19:53:34 <johill> ok, I think I got that, let me try
2007-04-21T20:11:28 <neagulm> moin
2007-04-21T20:11:41 <xorAxAx> hi neagulm
2007-04-21T20:43:16 <johill> hrm
2007-04-21T20:43:24 <johill> my server gets an error but other than that it probably works
2007-04-21T20:45:09 <johill> hm
2007-04-21T20:45:10 <johill> OpenID error: No matching endpoint found after discovering http://johannes:8081/Johannes.
2007-04-21T20:46:04 <johill> IDP support will have to wait, gotta go
2007-04-21T21:51:40 * johill is puzzled
2007-04-21T21:58:17 <johill> keturn: I can't make the idp work
2007-04-21T21:58:32 <johill> somehow the server is trying to find an association for http://localhost/|dumb while we only have some for http://localhost/|normal
2007-04-21T22:03:37 <johill> darn, doesn't myopenid.com provide an IDP on the frontpage?
2007-04-21T22:07:49 <johill> haha
2007-04-21T22:07:51 * johill gives up on that
2007-04-21T22:08:00 <johill> none of the providers listed on the openid.net wiki support that
2007-04-21T22:35:52 <CIA-28> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> * 2006:857118e9fdaa /docs/ (CHANGES CHANGES.fpletz): integrated CHANGES.fpletz into main CHANGES, TODOs/unclear stuff went to MoinMoin:XapianIntegration
2007-04-21T22:35:55 <CIA-28> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> * 2007:41ee59ba0b59 /MoinMoin/ (6 files in 5 dirs): merged main
2007-04-21T22:38:53 <CIA-28> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> * 2008:500eebb4da0f /docs/CHANGES.fpletz: removed CHANGES.fpletz
2007-04-21T22:40:45 <ThomasWaldmann> fpletz: please check last changesets and MoinMoin:XapianIntegration page
2007-04-21T23:42:42 <CIA-28> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> * 2009:f23bf37ffabf / (MoinMoin/xmlrpc/__init__.py docs/CHANGES): replace xmlrpc getUser by getUserProfile, auth can be done by getAuthToken/applyAuthToken in the same multicall MoinMoin: MoinMoinChat/Logs/moin-dev/2007-04-21 (last edited 2007-10-29 19:17:08 by localhost)
