MoinMoin: MoinMoinChat/Logs/moin-dev/2008-05-05

2008-05-05T08:21:53  <dreimark_> zenhase: take Xournal
2008-05-05T08:22:03  <dreimark_> moin
2008-05-05T08:50:36  <ThomasWaldmann> moin
2008-05-05T08:57:51  <mitsuhiko> ThomasWaldmann: ping
2008-05-05T08:58:55  <ThomasWaldmann> pong
2008-05-05T08:59:18  <zenhase> moin
2008-05-05T08:59:31  <mitsuhiko> hi zenhase
2008-05-05T08:59:42  <zenhase> hiho mitsuhiko :)
2008-05-05T08:59:47  <mitsuhiko> ThomasWaldmann: why have you asked about the wsgi iterator yesterday?
2008-05-05T09:00:01  <mitsuhiko> if there is something happening regarding wsgi i want to know :D
2008-05-05T09:00:15  <mitsuhiko> grml
2008-05-05T09:00:27  <ThomasWaldmann> i thought of a small optimization, using a list instead of StringIO
2008-05-05T09:00:52  <ThomasWaldmann> just was not sure if it is an optimization :)
2008-05-05T09:01:01  <mitsuhiko> i was hoping request.write would finally go away
2008-05-05T09:01:13  <ThomasWaldmann> if you say it is better to yield once, it is maybe not
2008-05-05T09:01:36  <zenhase> hmm
2008-05-05T09:01:51  <ThomasWaldmann> mitsuhiko: well, zenhase and you can make this possible, but request.write should still work
2008-05-05T09:02:22  <ThomasWaldmann> (doing global changes in a branch just makes a global merge problem)
2008-05-05T09:02:44  <mitsuhiko> ThomasWaldmann: so there won't be a template engine?
2008-05-05T09:02:47  <zenhase> removing all the write calls could be done after GSoC
2008-05-05T09:03:05  <mitsuhiko> and the DOM branch will not use request.write either i hope
2008-05-05T09:03:16  <ThomasWaldmann> mitsuhiko: we dont have a template engine in moin
2008-05-05T09:03:20  <zenhase> hope so to
2008-05-05T09:03:31  <zenhase> ThomasWaldmann: well .... ;)
2008-05-05T09:03:46  <dreimark> ....
2008-05-05T09:04:00  <ThomasWaldmann> zenhase: I am talking of production code :)
2008-05-05T09:04:05  <zenhase> hi dreimark
2008-05-05T09:04:35  <mitsuhiko> :-/
2008-05-05T09:04:55  * dreimark is after sleeping 12h wide awake
2008-05-05T09:05:28  <mitsuhiko> ThomasWaldmann: i just want to point out that request.write is somewhat of a problem for a wsgi environment and a terrible idea
2008-05-05T09:05:44  <mitsuhiko> imo the solution would be adding a compatibility layer so that the branches that depend on the request object get a new api
2008-05-05T09:05:48  <zenhase> especially in regard to the next wsgi spec
2008-05-05T09:05:52  <mitsuhiko> that is template engine, WSGI integration and DOM
2008-05-05T09:06:15  <zenhase> where the write-callable is about to be dropped
2008-05-05T09:06:39  <zenhase> hmm, wrapper object ...
2008-05-05T09:06:39  <mitsuhiko> zenhase: the write callable is a problem in the current wsgi version too as middlewares often break it
2008-05-05T09:06:55  <zenhase> mitsuhiko: yes, everybody disadvises to use it
2008-05-05T09:07:06  <zenhase> i also actually never saw it in production code
2008-05-05T09:07:38  <mitsuhiko> i saw it once in a application converted from a cgi script (hgweb)
2008-05-05T09:07:43  <zenhase> it's like many authors intuitively know, that generators are the way to go
2008-05-05T09:07:45  <mitsuhiko> but they are changing it
2008-05-05T09:08:02  <ThomasWaldmann> well, as I said, I am fine with offering an alternative to request.write, but request.write calls should not be touched and still work
2008-05-05T09:08:11  <mitsuhiko> ThomasWaldmann: how should that work?
2008-05-05T09:08:20  <ThomasWaldmann> ask xorAxAx :)
2008-05-05T09:08:25  <mitsuhiko> xorAxAx: how should that work?
2008-05-05T09:08:31  <zenhase> the alternative is returning string (or lists of strings ... or generators)
2008-05-05T09:08:46  <mitsuhiko> the sane solution would be getting rid of request.write
2008-05-05T09:08:52  <dreimark> mitsuhiko: later
2008-05-05T09:09:10  <dreimark> otherwise we have a very late 1.8 release period
2008-05-05T09:09:21  <dreimark> and 1.6 was too long
2008-05-05T09:09:26  <zenhase> mitsuhiko: actually this will require a lot more work than expected i think
2008-05-05T09:09:35  <zenhase> i have to admit that
2008-05-05T09:09:37  <mitsuhiko> not necessarily
2008-05-05T09:09:56  <zenhase> well, the request-object gets passed down deep into the moin-callstack
2008-05-05T09:10:11  <mitsuhiko> the only thing that has to change is that macros, parsers and the DOM are not allowed to use request.write
2008-05-05T09:10:17  <mitsuhiko> or request.write is a noop that writes into a buffer
2008-05-05T09:10:33  <zenhase> that's what i would propose
2008-05-05T09:11:48  <ThomasWaldmann> if you create a branch with merge problems in every file, consider the merging as part of your task :)
2008-05-05T09:12:06  <zenhase> it could be the app_iter which gets iterated on response(environ, start_response)
2008-05-05T09:12:18  <mitsuhiko> ThomasWaldmann: i'm not forcing you to drop request.write instantly. but the new apis should only use request.write in a wrapper, not natively
2008-05-05T09:12:41  <mitsuhiko> eg: the DOM writes into a buffer and once it's done the compatibility layer writes into request.write
2008-05-05T09:12:43  <zenhase> mitsuhiko: you mean the other way round?
2008-05-05T09:12:51  <mitsuhiko> so that switching can occur later easily
2008-05-05T09:12:59  <zenhase> a wrapper should provide request.write around the new api? :)
2008-05-05T09:13:01  <mitsuhiko> zenhase: no. the app iter must not include more than ~10 items
2008-05-05T09:13:05  <mitsuhiko> or it will become too slow
2008-05-05T09:13:11  <mitsuhiko> each yield does write() + flush()
2008-05-05T09:13:13  <zenhase> o_O?
2008-05-05T09:13:21  <mitsuhiko> yield is only there for chunked respones
2008-05-05T09:13:27  <mitsuhiko> streaming large files for example
2008-05-05T09:13:32  <zenhase> yes
2008-05-05T09:13:59  <zenhase> so it's a join before delivering the string
2008-05-05T09:14:05  <mitsuhiko> yes
2008-05-05T09:14:16  <zenhase> that's ok too of course
2008-05-05T09:16:26  <zenhase> hmm ... if we go for werkzeug as the wsgi-toolbox, we have request/response where as current moin has just a request, which also provides the api for the response (write and emit_headers e.g.)
2008-05-05T09:16:48  <mitsuhiko> zenhase: then merge it into one
2008-05-05T09:16:54  <mitsuhiko> or don't use Response
2008-05-05T09:17:12  <zenhase> well, the merging could be done in the wrapper object
2008-05-05T09:18:14  <zenhase> everybody not using the old API in the wrapper will get a reasonable Request (which we still have to design) and return output as strings
2008-05-05T09:18:15  <ThomasWaldmann> as far as the new dom stuff is concerned, i guess it is only 1 or a few request.write calls
2008-05-05T09:18:17  <zenhase> or Response
2008-05-05T09:18:49  <ThomasWaldmann> so it can be changed at the time we do that global change
2008-05-05T09:19:53  <zenhase> yeah. it would be obviously a request.write in .send_page, after the output has been generated
2008-05-05T09:20:50  <zenhase> mitsuhiko: how about the new API for the Request object (the real one, not the wrapper)?
2008-05-05T09:21:07  <zenhase> i have a list here, what i found to be stuck on the request-object in current moin
2008-05-05T09:21:17  <mitsuhiko> zenhase: http://paste.pocoo.org/show/47402
2008-05-05T09:21:49  <zenhase> yeah
2008-05-05T09:22:26  <mitsuhiko> zenhase: werkzeug does the encoding automatically, so you don't need the encoding in write
2008-05-05T09:22:40  <zenhase> i have things here like some strange dicts, pagemetadata, page, rootpage, editlog, user, actions ...
2008-05-05T09:22:56  <zenhase> erm
2008-05-05T09:22:57  <zenhase> action
2008-05-05T09:23:21  <zenhase> also i think those were sometimes a bit inconsistent
2008-05-05T09:23:30  <ThomasWaldmann> mitsuhiko: that's almost the same as we already have
2008-05-05T09:25:14  <zenhase> that are some of the attributes, moin-code expects on the request-object
2008-05-05T09:25:37  <zenhase> another list i have here contains the application code that's in the request-object ;)
2008-05-05T09:26:14  <zenhase> actually whole RequestBase.run looks like it should be an WSGI app
2008-05-05T09:26:22  <mitsuhiko> yes indeed
2008-05-05T09:26:34  <zenhase> and the supporting functions should become module-functions
2008-05-05T09:26:46  <zenhase> so request itself becomes more lightweight
2008-05-05T09:26:54  <mitsuhiko> not much
2008-05-05T09:27:01  <mitsuhiko> it's still the trashbin for everything
2008-05-05T09:27:17  <mitsuhiko> even the table of contents macro dumps it's intermediate data on that object
2008-05-05T09:27:20  <zenhase> well, that's what we want to change too, don't we?
2008-05-05T09:27:55  <ThomasWaldmann> zenhase: i suggest you do small steps and not try to turn all upside down at once
2008-05-05T09:27:58  <zenhase> perhaps it won't change in the first iteration, but we could fixate the goal we are working towards to ;)
2008-05-05T09:28:43  <zenhase> ThomasWaldmann: don't get me wrong ... i want to support all this on the compat wrapper, which would be first
2008-05-05T09:28:49  <ThomasWaldmann> one small step would be to change all the http stuff in request object to use wsgi stuff
2008-05-05T09:29:03  <zenhase> which would actually be mostly a copy of requestbase
2008-05-05T09:30:37  <zenhase> or let's say requestbase will be renamed to requestcompat and will inherit from the new slick request which in turn is based on werkzeug
2008-05-05T09:31:05  <ThomasWaldmann> another step is providing interfaces and docs for the current non-wsgi request methods
2008-05-05T09:31:15  <zenhase> then like you say, remove all http related stuff from requestbase and make it use the wsgi-stuff
2008-05-05T09:31:27  <zenhase> hmm
2008-05-05T09:32:04  <zenhase> yes ... so you have this huge compat-wrapper
2008-05-05T09:32:10  <zenhase> which is like current requestbase
2008-05-05T09:32:26  <zenhase> and then one can start moving functions from there into a module
2008-05-05T09:32:33  <zenhase> and make the wrapper use them
2008-05-05T09:32:46  <zenhase> so everything keeps working and working for the old code
2008-05-05T09:33:18  <zenhase> and you slowly build the new api, which new code can start using and which old code can be ported to
2008-05-05T09:33:47  <zenhase> does this plan sound ok?
2008-05-05T09:35:28  <ThomasWaldmann> that sounds more doable
2008-05-05T09:36:34  <mitsuhiko> zenhase: if the conversion takes place later yes
2008-05-05T09:36:46  <mitsuhiko> if the compat ends up being there for ages because nobody finishes it a huge -1
2008-05-05T09:36:51  <mitsuhiko> that would make the whole situation worse
2008-05-05T09:38:29  <ThomasWaldmann> well, you can see how far you get in SOC
2008-05-05T09:39:05  <ThomasWaldmann> but moin is too big and has too many extensions to do a quick complete rewrite
2008-05-05T09:42:36  <zenhase> mitsuhiko: no, i don't want to see the whole codebase rot on a compat layer :)
2008-05-05T09:43:25  <zenhase> mitsuhiko: if everything in this GSoC works as expected, next thing will be my other proposal i did
2008-05-05T09:43:47  <zenhase> which will eliminate a lot of the need for compat code
2008-05-05T09:46:11  <xorAxAx> mitsuhiko: my point was that supporting request.write in legacy code is easy
2008-05-05T09:46:33  <xorAxAx> mitsuhiko: request.write would simply append to a list and the invocation code would yield that list when necessary
2008-05-05T09:46:37  <xorAxAx> pretty trivial
2008-05-05T09:46:47  <zenhase> xorAxAx: see mitsuhikos paste up there
2008-05-05T09:46:47  <mitsuhiko> xorAxAx: ah. but that was not what i was talking about :)
2008-05-05T09:47:15  <xorAxAx> zenhase: thats unrelated
2008-05-05T09:47:24  <xorAxAx> mitsuhiko: but i was yesterday, pretty clearly :)
2008-05-05T09:47:34  <zenhase> hmm?
2008-05-05T09:47:46  <zenhase> why is it unrelated?
2008-05-05T09:48:58  <xorAxAx> zenhase: because i only see a mock class of a response object :)
2008-05-05T09:49:11  <xorAxAx> request even
2008-05-05T09:49:51  <xorAxAx> i guess i wouldnt even put the write stuff into the real request object
2008-05-05T09:50:01  <zenhase> xorAxAx: it's not a mock
2008-05-05T09:50:15  <xorAxAx> .... in my scenario .... but just into a proxy class that is used in legacy cases
2008-05-05T09:50:19  <xorAxAx> zenhase: so? :)
2008-05-05T09:50:20  <zenhase> it was just to show how this could be put on a werkzeug-request
2008-05-05T09:50:41  <zenhase> and that's essentially the same you would put it anywhere :)
2008-05-05T09:51:10  <xorAxAx> what do you mean by
2008-05-05T09:51:11  <xorAxAx> 09:50:41 < zenhase> and that's essentially the same you would put it anywhere :)
2008-05-05T09:51:14  <xorAxAx> ?
2008-05-05T09:51:17  <zenhase> xorAxAx: regarding your proxy class: see my plan on requestbase -> requestcompat
2008-05-05T09:51:39  <xorAxAx> zenhase: where is that plan? :
2008-05-05T09:51:44  <zenhase> xorAxAx: that it was just to demonstrate the simplicity of support
2008-05-05T09:51:55  <zenhase> xorAxAx: right now it's the discussion above
2008-05-05T09:52:06  <zenhase> i still have to put it on wiki
2008-05-05T09:52:18  <xorAxAx> i dont see why it would be huge
2008-05-05T09:52:31  <zenhase> well, because requestbase is huge?
2008-05-05T09:52:42  <xorAxAx> but nearly no of its methods are relevant for legacy code
2008-05-05T09:52:48  <xorAxAx> s/no/&ne/
2008-05-05T09:53:15  <xorAxAx> 80/20 rule, you know :)
2008-05-05T09:53:21  <zenhase> well, all the objects attached to it are
2008-05-05T09:53:29  <xorAxAx> which objects?
2008-05-05T09:53:43  <zenhase> page, rootpage, cfg, formatters, several dicts
2008-05-05T09:53:45  <zenhase> etc.
2008-05-05T09:54:38  <xorAxAx> formatters?
2008-05-05T09:54:55  <xorAxAx> those attrs dont make it huge code-wise :) thats one line of code each
2008-05-05T09:55:03  <xorAxAx> but maybe you meant attribute-wise
2008-05-05T09:56:36  <CIA-48> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3555:dc7590f9c3b0 1.7/MoinMoin/request/request_fcgi.py: request_fcgi.send_file(): flush often to avoid big memory consumption for sending of big file attachments
2008-05-05T09:58:46  <zenhase> xorAxAx: i never said it's huge codewise
2008-05-05T09:59:12  <zenhase> xorAxAx: it's huge as an object ... so much stuff collected there
2008-05-05T09:59:54  <CIA-48> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2641:cee6123099a6 1.6/MoinMoin/request/request_fcgi.py: request_fcgi.send_file(): flush often to avoid big memory consumption for sending of big file attachments
2008-05-05T10:01:09  <zenhase> and there is some stuff there which is not very 'requesty'
2008-05-05T10:01:35  <xorAxAx> zenhase: yeah
2008-05-05T10:01:45  <zenhase> like pagename normalization and theme init
2008-05-05T10:02:03  <zenhase> some 'make_unique_id' code i didn't quite get what it is for
2008-05-05T10:02:37  <zenhase> oh btw ... just realized something:
2008-05-05T10:02:49  <zenhase> surge protection -> middleware?
2008-05-05T10:03:04  <zenhase> it seems to me it can be plugged in into a WSGI stack
2008-05-05T10:03:15  <mitsuhiko> zenhase: -1 for that
2008-05-05T10:03:22  <zenhase> mitsuhiko: :o
2008-05-05T10:03:34  <mitsuhiko> at ubuntuusers we have such limits in the form validations, we came from middlewares
2008-05-05T10:03:36  <mitsuhiko> was a bad idea
2008-05-05T10:03:42  <zenhase> ah ok
2008-05-05T10:03:49  <mitsuhiko> because preview and abort shouldn't trigger it
2008-05-05T10:04:02  <zenhase> oh ... did not think about that :o
2008-05-05T10:04:09  <zenhase> ok ok ;)
2008-05-05T10:05:03  <zenhase> hmm
2008-05-05T10:06:28  <zenhase> but it could definitely be put elsewhere too
2008-05-05T10:08:08  <zenhase> s/elsewhere/in a module
2008-05-05T10:10:26  <ThomasWaldmann> zenhase: please also have in mind / make a plan about how to keep moin working most of the time of your work
2008-05-05T10:12:36  <zenhase> that's what the compat wrapper is for :)
2008-05-05T10:13:26  <ThomasWaldmann> dreimark: searchinresults does not make much sense as long as the basic search has lots of problems
2008-05-05T10:17:52  <dreimark> heh, plans for some future extension
2008-05-05T12:22:42  <dennda> ThomasWaldmann: Am I late with something? (Planning to start with coding next week)
2008-05-05T12:23:30  <dennda> s/coding/working on it/
2008-05-05T12:25:25  <lanius> hi
2008-05-05T12:25:48  <dennda> hi lanius :)
2008-05-05T12:32:40  <ThomasWaldmann> dennda: no, but doing some preps in advance is better than panicking at the end :)
2008-05-05T12:33:03  <dennda> ThomasWaldmann: I agree
2008-05-05T12:34:04  <dennda> there's an exam at the end of this week that I'm planning to tackle (I'd rather want to do that now than within the coding period)
2008-05-05T12:36:33  <dennda> johill: I need a picture of yours, preferably where you look angry with a whip in your hand. That'll help me motivate whenever I become tired :)
2008-05-05T12:44:56  <ThomasWaldmann> dennda: better ask LH for that picture :P
2008-05-05T12:45:27  <dennda> I am wondering whether that is a sexist comment :)
2008-05-05T12:47:20  <dreimark> ThomasWaldmann: where are the transclusion params described. do we have a mimetype one too? e.g. i do want to render a cgi script as python script
2008-05-05T12:49:05  <ThomasWaldmann> i don't think we have a force mimetype one
2008-05-05T12:49:29  * dreimark ok, do add a FR 
2008-05-05T13:08:58  <ThomasWaldmann> btw, we should have some b2 / rc1 soon
2008-05-05T13:12:08  * dreimark reboots
2008-05-05T13:22:02  <johill> dennda: haha
2008-05-05T13:22:12  <johill> I'll ask my g/f to take one... ;)
2008-05-05T13:22:46  <xorAxAx> johill: lol
2008-05-05T13:27:20  <dreimark> haha
2008-05-05T13:36:42  <dennda> johill: good
2008-05-05T13:38:51  <johill> for now you can photoshop this picture: johannes.sipsolutions.net/files/segelschein.jpg
2008-05-05T13:40:57  <xorAxAx> were you 14 when it was taken?
2008-05-05T13:41:34  <johill> I have no idea, but somehow I doubt the picture was recent at the time
2008-05-05T13:41:36  <johill> :)
2008-05-05T13:41:57  <johill> mind you I was (and still am I guess) small so it probably wasn't ancient either
2008-05-05T13:43:06  <dennda> that's not scaring me enough
2008-05-05T13:43:08  <xorAxAx> yeah, i noticed that :-)
2008-05-05T13:43:12  <xorAxAx> dennda: lol, indeed
2008-05-05T13:44:32  <johill> heh
2008-05-05T13:48:42  <TheSheep> any picture looks scary if you add a goatee
2008-05-05T13:50:05  <johill> that would *really* be scary. me at that age with a goatee
2008-05-05T13:52:52  <ThomasWaldmann> you could also add some evil sheep :D
2008-05-05T13:54:34  <TheSheep> atos.wmid.amu.edu.pl/~sheep/johill.jpg
2008-05-05T13:54:35  <TheSheep> :P
2008-05-05T13:55:40  <johill> hahaha
2008-05-05T13:56:35  <ThomasWaldmann> or have some Ballmer video undertitled with "if you fail for moin, you will have to work for him"
2008-05-05T13:57:35  <TheSheep> Wikizens! Wikizens! Wikizens!
2008-05-05T13:58:49  <ThomasWaldmann> http://moinmo.in/MoinMoinTodo/Release%201.7.0 < more ideas about xapian welcome
2008-05-05T13:59:15  <johill> ThomasWaldmann: I would suggest to disable it and do the search/backend storage merge I started
2008-05-05T13:59:16  <ThomasWaldmann> (esp. the rather SHORT term ones :)
2008-05-05T13:59:28  <johill> but I don't really have a good short-term idea
2008-05-05T13:59:31  <johill> yeah :)
2008-05-05T13:59:50  <xorAxAx> TheSheep: side parting  would complete the notion of an evil person
2008-05-05T14:00:19  <johill> o.O it's snowing (white petals or so)
2008-05-05T14:01:02  <TheSheep> xorAxAx: side parting?
2008-05-05T14:01:12  <xorAxAx> TheSheep: well, in the hairs
2008-05-05T14:01:19  * ThomasWaldmann somehow rather uses opera recently since hardy has ff3.0b5 :|
2008-05-05T14:02:07  <TheSheep> xorAxAx: ah, but isn't it illegal in Germany? ;)
2008-05-05T14:03:48  <xorAxAx> TheSheep: no
2008-05-05T14:24:31  <johill> yay finally figured a way to plot this
2008-05-05T14:24:32  <johill> http://johannes.sipsolutions.net/files/pb-dt.html
2008-05-05T14:24:58  <xorAxAx> whats that?
2008-05-05T14:25:33  <johill> my track from Thursday afternoon
2008-05-05T14:25:35  <xorAxAx> i was pretty near to that route when i rode to the extern stones :)
2008-05-05T14:25:43  <johill> I skated that way
2008-05-05T14:25:44  <xorAxAx> johill: by which device?
2008-05-05T14:25:47  <xorAxAx> ah
2008-05-05T14:25:52  <johill> inline skates
2008-05-05T14:26:04  <xorAxAx> how did you record the position?
2008-05-05T14:27:16  <johill> my N810 has GPS
2008-05-05T14:27:20  <dennda> there are special gps devices for that purpose
2008-05-05T14:28:49  <xorAxAx> johill: ah
2008-05-05T14:28:52  <xorAxAx> dennda: i know
2008-05-05T14:29:02  <xorAxAx> polar watches etc.
2008-05-05T14:33:46  <johill> fun, but going up Gauseköte and down again wasn't too hot, look at the elevation plot there :)
2008-05-05T14:37:37  <xorAxAx> how can i see it?
2008-05-05T14:38:02  <johill> need to upload it I guess
2008-05-05T14:38:18  <johill> unless you want to recreate it
2008-05-05T14:39:41  <johill> can you see this?
2008-05-05T14:39:44  <johill> http://johannes.sipsolutions.net/files/pb-dt.svg
2008-05-05T14:40:02  <johill> it's elevation over time
2008-05-05T14:40:38  <xorAxAx> yes
2008-05-05T14:40:58  <xorAxAx> wow
2008-05-05T14:41:08  <johill> almost 200m up and down
2008-05-05T14:42:20  <johill> it renders really slowly in my browser
2008-05-05T14:42:49  <xorAxAx> it needs ~ 5 sec. in opera 9
2008-05-05T14:43:07  <johill> oh. faster here, heh
2008-05-05T14:43:28  <xorAxAx> hmm, more like 3
2008-05-05T14:43:37  <johill> yeah
2008-05-05T14:43:58  * johill uploads .png
2008-05-05T14:44:33  <johill> smaller even
2008-05-05T14:45:31  <johill> (filesize I mean)
2008-05-05T17:53:49  <dreimark> http://crackle.com/c/Political/dust_devil/457363
2008-05-05T18:39:39  <zenhase> re
2008-05-05T20:17:42  <ThomasWaldmann> re
2008-05-05T20:39:02  <CIA-53> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3556:2b6f1d6c485d 1.7/MoinMoin/request/__init__.py: request.__init__: PEP8 whitespace fix
2008-05-05T20:39:30  * dreimark tries to write a test for the user / dict trouble
2008-05-05T22:10:50  <CIA-53> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3557:9052c06d4ae3 1.7/MoinMoin/_tests/ (__init__.py test_wikidicts.py): test_wikidicts: some more tests for group pages
2008-05-05T22:11:30  <dreimark> have not had luck to reproduce http://moinmo.in/MoinMoinBugs/17devWikiDictsDontUpdate
2008-05-05T22:16:14  <ThomasWaldmann> dreimark: happened on master17.moinmo.in
2008-05-05T22:16:56  <dreimark> how do we deploy mod_wsgi?
2008-05-05T22:17:13  <dreimark> ThomasWaldmann:
2008-05-05T22:17:51  <ThomasWaldmann> dpkg -i libapache2-mod-wsgi20.deb :)
2008-05-05T22:19:03  <dreimark> for group dicts it does not matter if a user is created or not
2008-05-05T22:19:10  <dreimark> it can be a page too
2008-05-05T22:19:51  <dreimark> the master group page is quite large
2008-05-05T22:23:44  <dreimark> ThomasWaldmann: can it be a problem caused from the reason for the simple_wrapper from server_wsgi too?
2008-05-05T22:26:37  <dreimark> htg
2008-05-05T22:29:26  <ThomasWaldmann> i dont understand your question
2008-05-05T23:04:03  <johill> how much in-emory caching does moni do now?
2008-05-05T23:04:15  <johill> it needs 100M memory here
2008-05-05T23:10:18  <LotekMarcel> on my apache with wsgi (only 2 cpus; it's consume around 50 - 120mb)
2008-05-05T23:10:28  <LotekMarcel> sorry, not cpus - process=2
2008-05-05T23:11:23  <LotekMarcel> but the cache going up slowly; so need to restart every day the apache (or using some maximum_requests)
2008-05-05T23:12:43  <LotekMarcel> I'm uising now some cron job to keep an eye on it http://www.heavy.ch/memory.log
2008-05-05T23:16:32  <ThomasWaldmann> johill: it caches quite some stuff, i usually see 10-120mb
2008-05-05T23:17:03  <ThomasWaldmann> (part if the process size might be related to the way python manages memory)
2008-05-05T23:24:48  <johill> hm