MoinMoin: MoinMoinChat/Logs/moin-dev/2007-07-14

2007-07-14T00:01:40  <neagulm> dreimark, ok
2007-07-14T00:02:27  <neagulm> dreimark, I'm going to sleep... I will write tomorrow the documentation for SpamBayes XML-RPC in order to allow early testing.
2007-07-14T00:04:12  <dreimark> neagulm: fine
2007-07-14T00:04:20  <dreimark> neagulm: good night
2007-07-14T00:04:39  <neagulm> good night
2007-07-14T00:40:34  <dreimark> xorAxAx: I do want general general classification capabilities too. orange seems not to have unicode support.
2007-07-14T00:50:13  <dreimark> moin ThomasWaldmann
2007-07-14T00:51:01  <dreimark> the test_sourcecode needs some improvement ;)
2007-07-14T00:52:47  <dreimark> it does not show an error if = or == is not surrounded by a blank, e.g. is_spam=True, category==self.spam_page
2007-07-14T00:59:34  <ThomasWaldmann> i just reused the pep8 code of someone else :)
2007-07-14T01:29:13  <dreimark> perhaps the comment blank should be examined too #body = body.encode("utf-8")
2007-07-14T01:37:05  <dreimark> gn
2007-07-14T01:47:51  <grzywacz> ThomasWaldmann, are you alive? 8)
2007-07-14T02:50:05  <grzywacz> I know nothing about xapian, but when page gets deleted, shouldn't its index get altered?
2007-07-14T02:50:59  <grzywacz> Ok, maybe saveText does the trick...
2007-07-14T02:52:19  <grzywacz> Also, looking at that function, I see there's an additional keyword argument involved in generation of index.
2007-07-14T02:52:54  <grzywacz> No idea how to handle it in case of generating indexes in an event handler...
2007-07-14T02:56:59  <grzywacz> Sure, it's 1 by default, and this case is easy, but about the other one? Is it even used?
2007-07-14T11:55:31  <ThomasWaldmann> moin
2007-07-14T12:38:52  <zenhase> moin
2007-07-14T12:46:57  <ThomasWaldmann> http://picasaweb.google.com/lenscape/Vilnius some pics from EuroPython
2007-07-14T13:01:08  <ThomasWaldmann> zenhase: btw, i have had a short chat with christopher lenz ( http://www.cmlenz.net/ ) about moin and genshi and he offered to help if you have questions
2007-07-14T13:01:34  <zenhase> ThomasWaldmann: i know, i already talked to him last night
2007-07-14T13:02:32  <ThomasWaldmann> ah, ok :)
2007-07-14T13:04:28  <zenhase> when i said i have to think about it again, i also talked to him about the proposals i had written down
2007-07-14T13:07:08  <zenhase> tho things might look messy, he advised me to take on request.write and the templates only in the first place
2007-07-14T13:08:51  <Fabi> I doubt there is time for anything more
2007-07-14T13:09:01  <zenhase> as you and xorAxAx said, this will keep me busy already for quite some time
2007-07-14T13:09:35  <zenhase> Fabi: yes, maybe so
2007-07-14T13:11:57  <Fabi> are there any objections against removing request.wrie from the code?
2007-07-14T13:13:05  <xorAxAx> yes
2007-07-14T13:13:21  <xorAxAx> IMHO it doesnt make any sense
2007-07-14T13:13:39  <xorAxAx> as long as there is no compatiblity stuff that does the capturing for old-school plugins
2007-07-14T13:13:53  <xorAxAx> its really not hard to achieve that and it would be a first step
2007-07-14T13:14:56  <Fabi> the current theme code does output capturing and of course it will have to stay there for a time
2007-07-14T13:15:19  <Fabi> the question is if the goal of removing request.write is the right one
2007-07-14T13:15:31  <xorAxAx> it doesnt gain much IMHO
2007-07-14T13:15:37  <Fabi> (sorry, don't have the log of last night here)
2007-07-14T13:15:56  <xorAxAx> except for lost streaming
2007-07-14T13:16:03  <xorAxAx> which is not such a bad issue IMHO
2007-07-14T13:16:23  <xorAxAx> hmm, no, streaming is not possible at all using genshi
2007-07-14T13:16:33  <xorAxAx> so it doesnt have any value at all
2007-07-14T13:16:41  <xorAxAx> zenhase: do you see any advantages?
2007-07-14T13:16:46  <zenhase> yes
2007-07-14T13:16:52  <xorAxAx> which ones?
2007-07-14T13:17:24  <Fabi> the question is if streaming really is a big win
2007-07-14T13:17:53  <Fabi> the mix of request.write and returning strings is a mess
2007-07-14T13:18:05  <zenhase> a more clear flow of data and the ability to hook in where needed because content is actually something returned, not written to the request around your back
2007-07-14T13:18:17  <xorAxAx> Fabi: no, streaming is not possible
2007-07-14T13:18:18  <Fabi> that forces an inorder call tree on the whole moin codes
2007-07-14T13:18:21  <zenhase> also i don't see why streaming would vanquish
2007-07-14T13:18:32  <xorAxAx> Fabi: in terms of avoiding to buffer data in the mem
2007-07-14T13:18:59  <zenhase> genshi does it all the time ... you apply lots of filters and at the end it's a single generator function yielding
2007-07-14T13:19:04  <xorAxAx> the problem of streaming is orthogonal
2007-07-14T13:19:13  <Fabi> as pages are small and caches are big I doubt that this matters much
2007-07-14T13:19:19  <xorAxAx> zenhase: the point is that genshi keeps all tags in memory
2007-07-14T13:19:23  <xorAxAx> zenhase: moin 1.5 doesnt
2007-07-14T13:19:29  <xorAxAx> it spits out small chunks
2007-07-14T13:19:31  <zenhase> xorAxAx: for the templates yes
2007-07-14T13:19:33  <ThomasWaldmann> i suggest staying away from the content parts of moin
2007-07-14T13:19:34  <xorAxAx> see
2007-07-14T13:19:45  <zenhase> but the generated output is streamlined to the user
2007-07-14T13:19:55  <xorAxAx> thats completly orthogonal to this proposed refactoring
2007-07-14T13:19:58  <zenhase> and well, it's really low memory
2007-07-14T13:20:03  <xorAxAx> zenhase: thats irrelevant
2007-07-14T13:20:09  <ThomasWaldmann> if you need the content stuff in some string and not directly output, then redirect the output done by request.write
2007-07-14T13:20:14  <xorAxAx> the memory usage is the same
2007-07-14T13:20:30  <xorAxAx> yeah, the stack of capturing works fine, zenhase
2007-07-14T13:20:40  <xorAxAx> and i dont see any large architectural flaw here
2007-07-14T13:21:09  <Fabi> xorAxAx, if we do output capturing by default there is no streaming, too...
2007-07-14T13:21:13  <xorAxAx> furthermore, send_page is orthogonal as well
2007-07-14T13:21:19  <xorAxAx> Fabi: yes, thats my point
2007-07-14T13:21:25  <xorAxAx> Fabi: thats what i mean by orthogonal
2007-07-14T13:21:34  <zenhase> xorAxAx: send_page is orthogonal in what sense?
2007-07-14T13:21:41  <xorAxAx> so please dont use invalid reasons to stall your development
2007-07-14T13:22:02  <Fabi> yes, but this is another hack ontop of a suboptimal design
2007-07-14T13:22:05  <xorAxAx> zenhase: that any refactoring idea presented in the last weeks does neither simplify or complicate refactoring of send_page
2007-07-14T13:22:16  <ThomasWaldmann> btw, one of our todos is emitting headers late (to make it possible to adapt the caching headers according to the type of content)
2007-07-14T13:22:34  <ThomasWaldmann> this doesnt go well with streaming, too
2007-07-14T13:22:40  <xorAxAx> zenhase: if you disagree, please explain why, on a wikipage.
2007-07-14T13:23:01  <zenhase> ThomasWaldmann: and also doesn't go well with request.write
2007-07-14T13:23:04  <xorAxAx> Fabi: what is suboptimal?
2007-07-14T13:23:27  <xorAxAx> zenhase: i disagree
2007-07-14T13:23:31  <ThomasWaldmann> zenhase: it does go well with r.w, if r.w isnt outputting to the browser, but is redirected
2007-07-14T13:24:00  <zenhase> so you want output-capturing hack to have more control over the headers?
2007-07-14T13:24:02  <Fabi> being forced to keep the call order and having a two different interfaces for generating output
2007-07-14T13:24:26  <xorAxAx> Fabi: what do you mean by 2 different interfaces? what do you mean by call order? the formatter interface?
2007-07-14T13:24:27  <ThomasWaldmann> zenhase: if you want to get things done and not end up with a huge workload and unfinished things, try to keep the amount of code you have to touch low
2007-07-14T13:24:45  <xorAxAx> Fabi: thats still something different or did anybody speak about changing formatter semantics?
2007-07-14T13:24:54  <Fabi> you ahve to make sure everything is called in the order it should apear in the page
2007-07-14T13:25:08  <xorAxAx> thats the formatter interface, its completly orthogonal to r.w
2007-07-14T13:25:20  <Fabi> and (saying it the 3rd time) being able to either return a string or write to r.w
2007-07-14T13:25:40  <xorAxAx> Fabi: i dont understand the last sentence
2007-07-14T13:26:40  <Fabi> macros for example are expected to return a string as there result (as parsers and formatters, too)
2007-07-14T13:26:53  <xorAxAx> parsers write to r.w
2007-07-14T13:27:07  <xorAxAx> yeah, whats your requirement here?
2007-07-14T13:27:09  <Fabi> but it is also kind of allowed to directly call r.w
2007-07-14T13:27:17  <Fabi> one interface
2007-07-14T13:27:18  <xorAxAx> macros may not call r.w
2007-07-14T13:27:26  <Fabi> some do
2007-07-14T13:27:32  <xorAxAx> which?
2007-07-14T13:27:49  <Fabi> RecentChanges to name a popular one
2007-07-14T13:28:03  <zenhase> Include
2007-07-14T13:28:08  <zenhase> (via .send_page)
2007-07-14T13:28:10  <xorAxAx> hmm
2007-07-14T13:28:44  <xorAxAx> and parsers can return strings?
2007-07-14T13:28:46  <xorAxAx> thats new to me
2007-07-14T13:28:53  <Fabi> hmm, let me see
2007-07-14T13:29:20  <xorAxAx> its pretty stupid to design some new architecture without having analysed the problem completly first
2007-07-14T13:29:27  <zenhase> no they don't but i think they should
2007-07-14T13:29:56  <xorAxAx> Fabi talked about the in-order issue
2007-07-14T13:30:22  <xorAxAx> of course you remove the issue a bit by requiring to return stuff ..
2007-07-14T13:30:40  <xorAxAx> hmm
2007-07-14T13:30:54  <xorAxAx> ok, what about this:
2007-07-14T13:31:29  <xorAxAx> macros may not use r.w anymore
2007-07-14T13:31:41  <xorAxAx> parsers are still allowed to use r.w, but only for legacy reasons
2007-07-14T13:31:51  <xorAxAx> the whole control flow passes strings
2007-07-14T13:32:03  <xorAxAx> --> parser calls are wrapped into output capturing
2007-07-14T13:32:18  <xorAxAx> still i think that this is orthogonal to theming issues :)
2007-07-14T13:33:10  <xorAxAx> Fabi: now explain how this should work together with the caching formatter
2007-07-14T13:33:32  <xorAxAx> the ordering issue is inherent ... you cant fix it by returning strings!
2007-07-14T13:33:46  <xorAxAx> imagine the docbook formatter, all methods dont return stuff at all (only "")
2007-07-14T13:34:11  <xorAxAx> you would need to redesign the pipeline completly
2007-07-14T13:34:21  <xorAxAx> and you are not the first to think about this, have a look into the wiki
2007-07-14T13:34:25  <xorAxAx> comments please
2007-07-14T13:34:27  <zenhase> it's not a pipeline at the moment
2007-07-14T13:34:38  <xorAxAx> well, call it differently :)
2007-07-14T13:34:39  <zenhase> that's the problem i am having with it
2007-07-14T13:34:52  <xorAxAx> chaos-engine
2007-07-14T13:34:56  <zenhase> you cannot easily get in there somewhere
2007-07-14T13:35:10  <xorAxAx> i dont want to talk about personal problems right now
2007-07-14T13:35:13  <zenhase> and put a "filter" in it
2007-07-14T13:35:20  <xorAxAx> please think about the issues and suggestions i brought up above
2007-07-14T13:35:38  <xorAxAx> do we have any use case for a filter?
2007-07-14T13:36:01  <zenhase> this was an allegory to the pipeline-image
2007-07-14T13:36:45  <xorAxAx> so, the above sketched change still doesnt free us from ordering issues but may support the simplification of send_page
2007-07-14T13:36:46  <zenhase> like somewhere at the lower end of the pipeline the template engine sitting there, waiting to receive everything it needs
2007-07-14T13:36:51  <Fabi> if everyone is ok with doing fine with exessive output capturing we can drop that r.w issue again
2007-07-14T13:36:55  <xorAxAx> is that what we want, or isnt it still enough?
2007-07-14T13:37:06  <Fabi> but the benefits of using r.w will go away anyway then
2007-07-14T13:37:15  <xorAxAx> Fabi: performance/memory-wise it doesnt make anything different
2007-07-14T13:37:40  <xorAxAx> because something has to capture it always
2007-07-14T13:37:46  <Fabi> yes, that's clear. I am talking about amintainability
2007-07-14T13:37:58  <xorAxAx> yes, what do you think about the suggestion above?
2007-07-14T13:38:28  <Fabi> capturing parsers?
2007-07-14T13:38:51  <xorAxAx> well, optionally of course, the wrapper would notice if it should return its buffer or the return value
2007-07-14T13:39:06  <xorAxAx> the idea would be that you could simplify a lot in the code with the above rules
2007-07-14T13:39:09  <xorAxAx> but still some work
2007-07-14T13:39:16  <Fabi> yes, sounds reasonabe, as we need to keep some capureing for legacy reasons anyway
2007-07-14T13:39:24  <xorAxAx> and compatiblity would still remain
2007-07-14T13:39:39  <xorAxAx> furthermore, it could be expanded to some sane non-string stream api lateron
2007-07-14T13:40:07  <xorAxAx> zenhase: thoughts? :)
2007-07-14T13:40:27  <Fabi> there is a much more interesting problem about r.w than the parsers: actions
2007-07-14T13:41:16  <xorAxAx> yes, has anybody thought about it and wrote up suggestions?
2007-07-14T13:41:25  <Fabi> there currently is no standard what actions can/can't do at all
2007-07-14T13:42:00  <xorAxAx> they only can call r.w :)
2007-07-14T13:42:05  <xorAxAx> returning stuff would be useless
2007-07-14T13:42:13  <Fabi> I think it is impossible to fix that in a backward compatible way
2007-07-14T13:42:14  <zenhase> xorAxAx: why?
2007-07-14T13:42:20  <xorAxAx> zenhase: hmm?
2007-07-14T13:42:38  <zenhase> xorAxAx: why would it be useless?
2007-07-14T13:42:47  <xorAxAx> zenhase: because the return value is ignored currently
2007-07-14T13:42:51  <Fabi> of course we can keep our compat theme in place to keep things working
2007-07-14T13:43:12  <Fabi> but it would be better to move to a new interface there and move action per action
2007-07-14T13:43:16  <xorAxAx> well, i dont see a problem offering some compatiblity mode actions
2007-07-14T13:43:22  <xorAxAx> it would not be very complex
2007-07-14T13:43:34  <xorAxAx> s/mode/& for/
2007-07-14T13:45:34  <xorAxAx> so, it makes sense to do those transitions
2007-07-14T13:45:53  <xorAxAx> the only design work that needs to be done is thinking about an action compatilbity thingie
2007-07-14T13:46:10  <Fabi> ok, we'll discuss that and come back with the next iteration
2007-07-14T13:46:10  <xorAxAx> zenhase started with some suggestions to change it completly, i havent had a look at them
2007-07-14T13:46:38  <zenhase> o_O
2007-07-14T13:46:54  <xorAxAx> zenhase: i think your ideas went into the direction of implementing semantics geared towards item centric dispatching
2007-07-14T13:47:00  <xorAxAx> something like that is necessary, yes
2007-07-14T13:47:12  <grzywacz> moin
2007-07-14T13:47:28  <xorAxAx> moin grzywacz
2007-07-14T13:47:46  <xorAxAx> zenhase: so do you have any questions about send_page/parsers/macros?
2007-07-14T13:48:31  <Fabi> xorAxAx, "how to kill sendpage?"
2007-07-14T13:48:40  <zenhase> yeah, that's a nice one
2007-07-14T13:49:01  <xorAxAx> whats the exact issue about it?
2007-07-14T13:49:12  <xorAxAx> the above changes should simplify it
2007-07-14T13:49:28  <xorAxAx> you need to decouple send_page into a content-only part and a rest-part
2007-07-14T13:49:35  <xorAxAx> and then you start working from that onwards
2007-07-14T13:50:27  <zenhase> think anyone else tried that in the past 5 years and succeeded?
2007-07-14T13:51:11  <xorAxAx> i dont know about anybody who attempted it, n
2007-07-14T13:51:11  <xorAxAx> o
2007-07-14T13:51:32  <zenhase> well, i assumed that someone would have tried from time to time
2007-07-14T13:51:43  <xorAxAx> cant remember that, no
2007-07-14T13:51:53  <xorAxAx> i joined moin in autumn 2004
2007-07-14T13:52:35  <xorAxAx> so can you describe your problems you had while doing so?
2007-07-14T13:52:55  <xorAxAx> did you attempt the 2-method goal i described a few lines above?
2007-07-14T13:53:51  <zenhase> did you read the wikipages?
2007-07-14T13:53:57  <xorAxAx> which one?
2007-07-14T13:54:16  <zenhase> DesignPaper
2007-07-14T13:54:20  <zenhase> the part one send_page
2007-07-14T13:54:24  <zenhase> s/one/on
2007-07-14T13:57:30  <xorAxAx> i dont see any answer to my question
2007-07-14T13:59:40  <zenhase> xorAxAx: which one?
2007-07-14T13:59:53  <xorAxAx> 13:52:54 < xorAxAx> did you attempt the 2-method goal i described a few lines above?
2007-07-14T14:00:16  <zenhase> well, i was referring to your first question
2007-07-14T14:00:24  <xorAxAx> ah, ok
2007-07-14T14:00:29  <xorAxAx> well
2007-07-14T14:00:32  <xorAxAx> its very general
2007-07-14T14:00:38  <xorAxAx> it says "its too complex, i cant cope"
2007-07-14T14:00:47  <xorAxAx> so you need to divide the problem into small step
2007-07-14T14:00:48  <xorAxAx> s
2007-07-14T14:01:33  <zenhase> i really tried
2007-07-14T14:01:40  <zenhase> dividing that is
2007-07-14T14:01:52  <zenhase> that's what i didn't succeed at
2007-07-14T14:02:16  <xorAxAx> well, we all know that it is not impossible
2007-07-14T14:02:28  <xorAxAx> did you describe your new goal somewhere of that divided scenario?
2007-07-14T14:02:47  <xorAxAx> the way it should to be (at least in the first iteration)
2007-07-14T14:03:03  <Fabi> xorAxAx, the problem is to refactor send_page we need to change the actions
2007-07-14T14:03:17  <zenhase> and several other stuff
2007-07-14T14:03:32  <Fabi> including the editors
2007-07-14T14:03:39  <xorAxAx> Fabi: not if the compatilbity code is in place
2007-07-14T14:03:54  <xorAxAx> so dont worry about that
2007-07-14T14:04:01  <xorAxAx> and sketch some new architecture
2007-07-14T14:04:08  <Fabi> if you don't change the send_page interface there is not much to win
2007-07-14T14:04:10  <xorAxAx> then think about how to handle 80% of the old interfaces
2007-07-14T14:04:11  <zenhase> that's my overall problem ... you touch one thing, you have to touch many things ... you put compat-wrappers in place, your adding to the overall complexity
2007-07-14T14:04:19  <xorAxAx> Fabi: of course that one should be simplified
2007-07-14T14:04:43  <xorAxAx> zenhase: not really, it should become more simple
2007-07-14T14:04:50  <xorAxAx> in terms of the problems you are describing
2007-07-14T14:04:55  <xorAxAx> maybe more code, yes
2007-07-14T14:05:37  <zenhase> more code equals more complexity to me
2007-07-14T14:05:54  <xorAxAx> so you disregard the storage refacotring by lanius?
2007-07-14T14:06:05  <xorAxAx> i think he is writing a lot of code
2007-07-14T14:06:23  <zenhase> nope
2007-07-14T14:06:43  <zenhase> he did not touch much of other parts of moin yet
2007-07-14T14:06:51  <xorAxAx> one thing i dislike about your redesign page is that you suggest approaches that will change whole moin
2007-07-14T14:07:01  <xorAxAx> thats simply not possible because its tooo much work
2007-07-14T14:07:08  <zenhase> but i think when he does so, the outcome should be less code in the sum
2007-07-14T14:07:18  <xorAxAx> zenhase: no, it will be more
2007-07-14T14:07:52  <zenhase> because of compat hacks/
2007-07-14T14:07:53  <zenhase> ?
2007-07-14T14:07:55  <Fabi> I don't think we will win anything discussing other SOC projects here
2007-07-14T14:08:01  <xorAxAx> if you want to touch the formatter interface, then you should realise that it is a whole topic of its own
2007-07-14T14:08:11  <xorAxAx> zenhase: no, because its better partioned into functions etc.
2007-07-14T14:08:19  <zenhase> xorAxAx: ok, lets skip the formatter
2007-07-14T14:08:21  <xorAxAx> Fabi: i am just showing him that this assumption is incorrect
2007-07-14T14:08:34  <xorAxAx> zenhase: well, this means that you have to skip every filter and pipeline idea
2007-07-14T14:08:36  <xorAxAx> forget them :)
2007-07-14T14:09:32  <zenhase> i will take a break now
2007-07-14T14:09:37  <xorAxAx> and yes, some MVC-like stuff might be the goal - but its necessary to design something that is doable in 6 weeks and doesnt leave moin with a lot of broken edges
2007-07-14T14:10:01  <zenhase> hmm
2007-07-14T14:10:17  <xorAxAx> one tiny bit is the refactoring i described above
2007-07-14T14:10:22  <zenhase> can i ask you: why does it have to be doable in 6 weeks?
2007-07-14T14:10:29  <xorAxAx> i think you can tackle that. afterwards, we can think about the overall goal
2007-07-14T14:10:49  <zenhase> but wait
2007-07-14T14:10:52  <xorAxAx> zenhase: because you are judged for the stuff you will be doing in the remaining time frame. of course you can complete it afterwards
2007-07-14T14:11:28  <xorAxAx> in fact, we are encouraging students like you to try to stay in the team and try to continue contributing
2007-07-14T14:11:43  <xorAxAx> unfortunately, this hasnt worked out last year for anybody of the students
2007-07-14T14:12:22  <Fabi> xorAxAx, jugding is still my job - still for what ever reason
2007-07-14T14:12:37  <zenhase> i really need a break right now
2007-07-14T14:12:44  <xorAxAx> Fabi: yes, thats right - so do you aim for something unfinished that he could finish afterwards?
2007-07-14T14:12:49  <xorAxAx> zenhase: perfectly fine
2007-07-14T14:13:07  <Fabi> xorAxAx, nope
2007-07-14T14:13:36  <zenhase> bbl
2007-07-14T14:17:04  * xorAxAx generates a diffstat for lanius' branch
2007-07-14T14:17:17  <xorAxAx> hmm
2007-07-14T14:18:05  <xorAxAx>  34 files changed, 2952 insertions(+), 1226 deletions(-)
2007-07-14T14:18:20  <xorAxAx> so my intuition didnt got me wrong
2007-07-14T14:18:28  <xorAxAx> yes, he will still remove some stuff in the future
2007-07-14T14:18:38  <xorAxAx> but i am not sure if it will be as much as that
2007-07-14T14:21:02  <starshine> zenhase: it may helpt to step out of the code for a bit and look only at the pieces of design as you might describe them to someone who doesn't code.  a friend who's curious, a manager, a precocious child, a diagram.
2007-07-14T14:22:25  <starshine> in such elements you may find that some are the same, and consider if combining them would reduce complexity, or if they only use a similar Pattern or structure.
2007-07-14T14:23:31  <starshine> but most people take more than a few bites to finish a meal; it's cool to take things in smaller chunks.
2007-07-14T16:37:35  <grzywacz> xorAxAx, do you happen to know if disabling indexing in PageEditor (in saveText) is actually used?
2007-07-14T16:37:41  <grzywacz> *disabling of
2007-07-14T16:38:29  * xorAxAx checks the code
2007-07-14T16:39:35  <xorAxAx> grzywacz: line number?
2007-07-14T16:39:51  <grzywacz> xorAxAx, can't tell, I have some local changes. 8)
2007-07-14T16:39:55  <grzywacz> xorAxAx, search for xapian.
2007-07-14T16:39:58  <grzywacz> In saveText
2007-07-14T16:40:07  <grzywacz> It uses a kw argument to optionally disable indexing.
2007-07-14T16:40:20  <xorAxAx> you mean kw.get('index', 1)
2007-07-14T16:40:32  <grzywacz> Yes.
2007-07-14T16:40:39  <xorAxAx> no idea
2007-07-14T16:40:42  <xorAxAx> grep for index=
2007-07-14T16:40:45  <grzywacz> %)
2007-07-14T16:40:54  <xorAxAx> :)
2007-07-14T16:40:59  <johill> better index\s*= ;)
2007-07-14T16:41:01  <xorAxAx> yes, 2 usages
2007-07-14T16:41:01  <grzywacz> I'm asking because this can't currently be easily passed inside an event.
2007-07-14T16:41:07  <grzywacz> gh
2007-07-14T16:41:31  <xorAxAx> well
2007-07-14T16:41:53  <xorAxAx> both in pageditor
2007-07-14T16:42:13  <xorAxAx> copy and rename dont reindex
2007-07-14T16:42:15  <xorAxAx> no idea why
2007-07-14T16:42:37  <xorAxAx> ah, because they have to delete the old item
2007-07-14T16:42:47  <xorAxAx> see, so no problem for you
2007-07-14T16:43:05  <xorAxAx> you wont get any edit event in case of deletions/renames
2007-07-14T16:44:06  <grzywacz> Good. :)
2007-07-14T16:53:26  <xorAxAx> grzywacz: can you assert in send_event that the returned value is an instance of the appropriate base class?
2007-07-14T16:56:44  <grzywacz> Yes, good idea.
2007-07-14T16:59:55  <grzywacz> http://kolos.math.uni.lodz.pl/~grzywacz/lol.png
2007-07-14T16:59:56  <grzywacz> Failure. ;>
2007-07-14T17:00:28  <xorAxAx> hmm? :)
2007-07-14T17:00:52  <grzywacz> Size of the code window. ;>
2007-07-14T17:00:56  <xorAxAx> well
2007-07-14T17:01:04  <xorAxAx> you can put the outline below the navigator
2007-07-14T17:01:21  <xorAxAx> and move the console/problems view to the search view
2007-07-14T17:01:23  <grzywacz> I can maximize the code view as well, but it's suboptimal. x)
2007-07-14T17:07:12  <zenhase> re
2007-07-14T17:07:45  <grzywacz> re zenhase :)
2007-07-14T17:09:13  <grzywacz> I guess I should test my xapian changes before pushing...
2007-07-14T17:10:57  <xorAxAx> would be appreciated
2007-07-14T17:11:30  <grzywacz> Is there any page describing how to enable xapian in moin?
2007-07-14T17:11:37  <xorAxAx> yes
2007-07-14T17:11:39  <xorAxAx> some help page
2007-07-14T17:11:43  <xorAxAx> search for xapian
2007-07-14T17:11:54  <grzywacz> Did so, nothing comes up?
2007-07-14T17:11:57  <grzywacz> http://moinmo.in/?action=fullsearch&context=180&value=xapian&titlesearch=Tytu%C5%82y
2007-07-14T17:12:38  <xorAxAx> titlesearch?
2007-07-14T17:12:50  <grzywacz> ?
2007-07-14T17:13:12  <xorAxAx> you pressed titlesearch
2007-07-14T17:13:19  <xorAxAx> you have to click "text"
2007-07-14T17:13:58  <grzywacz> Did as well.
2007-07-14T17:16:25  <xorAxAx> you have to search locally
2007-07-14T17:16:34  <xorAxAx> not in moinmo.in
2007-07-14T17:18:32  <grzywacz> Your search query "xapian" didn't return any results. %)
2007-07-14T17:19:31  <xorAxAx> thats weird
2007-07-14T17:19:58  <xorAxAx> URL?
2007-07-14T17:20:13  <xorAxAx> there is even a help page called "HelpOnXapian"
2007-07-14T17:20:17  <xorAxAx> that is in the default underlay
2007-07-14T17:20:26  <xorAxAx> are you sure that you are running with underlay?
2007-07-14T17:20:50  <grzywacz> Well, I have other help pages available, so yes.
2007-07-14T17:21:39  <xorAxAx> on test.www.de i am seeing HelpOnXapian
2007-07-14T17:21:45  <xorAxAx> maybe you can find out why it is missing locally
2007-07-14T17:21:59  <grzywacz> There's no HelpOnXapian in my underlay.
2007-07-14T17:23:15  <grzywacz> Maybe it's missing in 1.7?
2007-07-14T17:24:11  <xorAxAx> ThomasWaldmann:
2007-07-14T17:24:27  <xorAxAx> btw, at someday we should underlay.tar.bz2 to be an underlay.tar file
2007-07-14T17:24:36  <xorAxAx> to reduce the growage of the repo
2007-07-14T17:51:28  <grzywacz> xorAxAx, help, "moin" tool mentioned in HelpOnXapian, is it the one located in wiki/server?
2007-07-14T17:53:20  <grzywacz> Hm, I guess not. :S
2007-07-14T17:53:42  <xorAxAx> nope :)
2007-07-14T17:53:52  <xorAxAx> its the one in MoinMoin.scripts
2007-07-14T17:54:04  <xorAxAx> well, a stupid idea to call both moin, indeed
2007-07-14T17:54:27  <grzywacz> All three, actually.
2007-07-14T17:54:31  <grzywacz> There's one in the root dir as well.
2007-07-14T17:56:01  <xorAxAx> hehe
2007-07-14T17:56:04  <xorAxAx> even worse
2007-07-14T17:56:57  <grzywacz> Also, docs mention "moin", and it's "moin.py" %)
2007-07-14T17:57:05  <grzywacz> (There's a bare "moin" as well %))
2007-07-14T17:57:12  <xorAxAx> you may fix that on moinmaster :)
2007-07-14T17:57:17  <xorAxAx> well
2007-07-14T17:57:24  <xorAxAx> setup.py might generate a moin file
2007-07-14T17:57:29  <xorAxAx> so maybe its right
2007-07-14T17:57:32  <xorAxAx> no idea
2007-07-14T17:57:33  <grzywacz> Hm, ok.
2007-07-14T17:58:37  * grzywacz waits for index to get generated
2007-07-14T18:06:33  <grzywacz> Owww, bug.
2007-07-14T18:08:17  <xorAxAx> where?
2007-07-14T18:09:06  <grzywacz> Trying to reproduce. 8)
2007-07-14T18:09:35  <grzywacz> Right it looks like like I'm killing the official wiki, because it's been processing my request for a minute or two O_o;
2007-07-14T18:10:18  <grzywacz> http://moinmo.in/%E7%BB%B4%E5%9F%BA%E6%B2%99%E7%9B%98%E6%BC%94%E7%BB%83?action=info&hitcounts=1
2007-07-14T18:10:22  <grzywacz> This link works weird.
2007-07-14T18:10:30  <grzywacz> It crashes in my local wiki.
2007-07-14T18:10:55  <grzywacz> (with "len() of unsized object")
2007-07-14T18:11:34  <grzywacz> (That's the page which comes up if you do a fulltext search on "asd" ;>)
2007-07-14T18:12:35  <xorAxAx> :-)
2007-07-14T18:26:34  <grzywacz> Ok, xapian seems to work as it is.
2007-07-14T18:27:44  <xorAxAx> did you also modify deletePage and renamePage?
2007-07-14T18:28:16  <grzywacz> Yes, I send events from there and moved the xapian code handling these cases into event handlers.
2007-07-14T18:29:04  <xorAxAx> ok
2007-07-14T18:29:33  <xorAxAx> maybe you find other places in the code that could use the events framework :)
2007-07-14T18:45:17  <grzywacz> 3 tests fail in 1.7, my fault or something global?
2007-07-14T18:48:31  <xorAxAx> well, you should have a main repo locally where you can execute py.test
2007-07-14T18:52:05  <xorAxAx> 3 failures in main, yes
2007-07-14T18:54:51  <xorAxAx> ThomasWaldmann: there are failing wikiutil tests
2007-07-14T18:55:56  <grzywacz> Ok, so I'll push my stuff with no worries. 8)
2007-07-14T18:56:21  <grzywacz> (spam)
2007-07-14T18:58:22  <xorAxAx> where?
2007-07-14T19:00:51  <grzywacz> <_<
2007-07-14T19:00:56  * grzywacz summons spam O_o;
2007-07-14T19:01:01  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2363:a73c99c076c2 1.7-jabber-knowak/MoinMoin/userprefs/suid.py: clean up suid prefs
2007-07-14T19:01:05  <grzywacz> Ah. 8)
2007-07-14T19:01:10  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2364:96ec23afac7d 1.7-jabber-knowak/MoinMoin/userprefs/oid.py: fix some bugs in openid prefs
2007-07-14T19:01:16  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2365:b9feee61d28e 1.7-jabber-knowak/MoinMoin/ (action/newaccount.py userprefs/prefs.py): split newaccount form from prefs
2007-07-14T19:01:25  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2366:2f36b737022a 1.7-jabber-knowak/MoinMoin/ (action/recoverpass.py userprefs/prefs.py): split recoverpass form from prefs
2007-07-14T19:01:32  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2367:405032c82324 1.7-jabber-knowak/MoinMoin/ (3 files in 2 dirs): introduce changepass userprefs plugin
2007-07-14T19:01:37  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2368:5fc0717a060f 1.7-jabber-knowak/MoinMoin/userprefs/ (__init__.py changepass.py prefs.py suid.py): refactor userprefs for more consistent UI
2007-07-14T19:01:44  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2369:ccf996479233 1.7-jabber-knowak/MoinMoin/ (request/__init__.py userprefs/prefs.py userprefs/suid.py): fix a suid bug: force switched-to user valid
2007-07-14T19:01:45  <grzywacz> CIA is slow today.
2007-07-14T19:01:49  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2370:6d9970321d79 1.7-jabber-knowak/MoinMoin/userprefs/oid.py: disable openid preferences if openid libs are not installed
2007-07-14T19:01:58  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2371:7de0a5eebf76 1.7-jabber-knowak/MoinMoin/parser/_tests/test_text_moin_wiki.py: first simple test for link markup, fails
2007-07-14T19:02:06  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2372:704c1051e73f 1.7-jabber-knowak/MoinMoin/parser/_tests/test_text_moin_wiki.py: more link tests as I expect things to behave
2007-07-14T19:02:11  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2373:7e7560f2cddd 1.7-jabber-knowak/MoinMoin/parser/_tests/test_text_moin_wiki.py: more tests for linking
2007-07-14T19:02:18  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2374:629dea07759f 1.7-jabber-knowak/MoinMoin/ (parser/text_moin_wiki.py wikiutil.py):
2007-07-14T19:02:21  <CIA-27> moin: do not parse single-quoted page names as quoted
2007-07-14T19:02:23  <CIA-27> moin: fixes bugs with ['text_html'] being a link
2007-07-14T19:02:25  <CIA-27> moin: Johannes Berg <johannes AT sipsolutions DOT net> * 2375:f9b0d11a6569 1.7-jabber-knowak/MoinMoin/parser/_tests/test_text_moin_wiki.py: use assert in linking test
2007-07-14T19:02:30  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2376:590caeeca03a 1.7-jabber-knowak/docs/CHANGES: Mention notification features in CHANGES.
2007-07-14T19:02:33  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2377:2b31b87db417 1.7-jabber-knowak/MoinMoin/events/emailnotify.py: Make name pep8 compatible.
2007-07-14T19:02:36  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2378:53fbdf5899ff 1.7-jabber-knowak/MoinMoin/ (PageEditor.py events/__init__.py events/jabbernotify.py): Use the event handling framework to (re)generate xapian indexes.
2007-07-14T19:02:44  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2379:aad0a118e609 1.7-jabber-knowak/MoinMoin/ (3 files in 3 dirs): Automatic whitespace fixes (?).
2007-07-14T19:02:47  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2380:06fabac5a1c7 1.7-jabber-knowak/ (7 files in 5 dirs): Merge main.
2007-07-14T19:06:20  <xorAxAx> grzywacz: your commit wasnt clean :)
2007-07-14T19:06:28  <xorAxAx> wrt MoinMoin/events/jabbernotify.py
2007-07-14T19:08:58  <grzywacz> xorAxAx, fixed a bug exposed by testing of xapian changes
2007-07-14T19:09:33  <xorAxAx> yes, i figured so
2007-07-14T19:09:49  <xorAxAx> nevertheless its unrelated :)
2007-07-14T19:11:11  <xorAxAx> ThomasWaldmann: ah, it looks like 629dea07759f did it
2007-07-14T19:12:45  <grzywacz> Omg...
2007-07-14T19:13:20  <grzywacz> I failed.
2007-07-14T19:13:34  <xorAxAx> hmm?
2007-07-14T19:13:40  <xorAxAx> in the banking exams?
2007-07-14T19:13:59  <grzywacz> No, in commiting.
2007-07-14T19:14:00  <grzywacz> xD
2007-07-14T19:14:00  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2381:ca82098b3e5f 1.7-jabber-knowak/MoinMoin/events/xapian_index.py: Add a file forgotten from the previous commits...
2007-07-14T19:14:07  <CIA-27> moin: Karol 'grzywacz' Nowak <grzywacz@sul.uni.lodz.pl> * 2382:abe038cca0c5 1.7-jabber-knowak/MoinMoin/events/xapian_index.py: Merge devel.
2007-07-14T19:14:17  <xorAxAx> hehe
2007-07-14T19:19:17  <xorAxAx> johill: 629dea07759f breaks 3 test (This message has been postponed.)
2007-07-14T19:19:43  <johill> uh huh, got kicked off somehow
2007-07-14T19:20:08  <johill> it *breaks* tests. hm
2007-07-14T19:20:54  <johill> xorAxAx: which ones?
2007-07-14T19:21:05  <xorAxAx> well, run py.test on all :)
2007-07-14T19:21:08  <xorAxAx> you will see
2007-07-14T19:21:15  <johill> it's taking a while :)
2007-07-14T19:21:19  <xorAxAx> (actually, those are wikiutil tests
2007-07-14T19:21:44  <xorAxAx> yes, i suggest moving that damn source code test into a subdirectory or depending on a config option to run it
2007-07-14T19:22:38  <johill> ah
2007-07-14T19:22:39  <johill> well
2007-07-14T19:22:49  <johill> the wikiutil tests explicitly tests ' quoting which I think is bogus to start with
2007-07-14T19:23:42  <johill> uh huh
2007-07-14T19:23:45  <johill> stupid pep8 test
2007-07-14T19:23:48  <johill> /home/johannes/Projects/moin/moin-1.7/MoinMoin/script/migration/linkconv.py:10:28: E203 whitespace before ':'
2007-07-14T19:23:51  <johill> i18n.wikiLanguages = lambda :[]
2007-07-14T19:23:54  <johill> well...
2007-07-14T19:24:09  <johill> /home/johannes/Projects/moin/moin-1.7/MoinMoin/script/migration/linkconv.py:297:20: W291 trailing whitespace
2007-07-14T19:24:18  <johill> yeah duh, it's a string and I want it there
2007-07-14T19:24:23  <johill> /home/johannes/Projects/moin/moin-1.7/MoinMoin/script/migration/linkconv.py:304:1: W191 indentation contains tabs
2007-07-14T19:24:23  <xorAxAx> hah
2007-07-14T19:24:26  <johill> same
2007-07-14T19:24:39  <johill> /home/johannes/Projects/moin/moin-1.7/MoinMoin/script/migration/linkconv.py:304:1: E101 indentation contains mixed spaces and tabs
2007-07-14T19:24:42  <johill> same
2007-07-14T19:24:44  <johill> crap
2007-07-14T19:25:19  <johill> "Sarah O'Connor"
2007-07-14T19:25:20  <johill> heh
2007-07-14T19:25:40  <johill> slight mix-up there?
2007-07-14T19:26:08  * johill ponders removing the broken tests there
2007-07-14T19:27:14  <johill> ThomasWaldmann: can you explain the situation with single-quote quoting?
2007-07-14T19:27:32  <johill> it breaks a lot and doesn't seem to gain us anything except the dubious ability to quote double-quotes
2007-07-14T19:27:45  <johill> (but not strings with mixed single/double quotes anyway)
2007-07-14T20:07:17  <vpv> do you think that the storage refactoring stuff could eventually solve some of the performance problems that the Fedora Wiki has?
2007-07-14T20:07:44  <vpv> if they could use a database backend, for example
2007-07-14T20:07:59  <xorAxAx> vpv: nope
2007-07-14T20:08:29  <xorAxAx> the tasks necessary for optimising that bottleneck are orthogonal
2007-07-14T20:11:15  <vpv> they just have a really long term plan of maybe someday replacing the wiki if the problems are bad enough, but personally I'd prefer if they could try to put some time into optimizing Moin. The Fedora Docs project is going to need Moin anyway, so...
2007-07-14T20:15:40  <xorAxAx> well, the necessary optimisation is done in less than 2 hours
2007-07-14T20:15:50  <xorAxAx> i think migration of any kind is much more costly
2007-07-14T20:16:47  <vpv> I don't really know the problems they have so well, but I agree, migration would be a real pain
2007-07-14T20:17:26  <xorAxAx> the only problem i know about is generating the list of people that are subscribed to a particular page
2007-07-14T20:18:06  <vpv> yes, I think that's it, that's what slows down the edits, other than that it's working pretty well
2007-07-14T20:18:15  <xorAxAx> yep
2007-07-14T20:19:43  <xorAxAx> basically, after every user save, some lists need to be updated that save for every page what the subscribers are
2007-07-14T20:21:59  <xorAxAx> using the new events framework, thats clean and easy to do
2007-07-14T21:36:18  <grzywacz> ?:)
2007-07-14T21:37:28  <xorAxAx> grzywacz: hmm? :)
2007-07-14T21:37:41  <grzywacz> I hear something about events. ;)
2007-07-14T21:39:46  <xorAxAx> yeah, fedora prefers a migration over some minimal performance fixes
2007-07-14T21:41:37  <grzywacz> No wonder if they are minimal, but still not done. ;]
2007-07-14T21:43:52  <xorAxAx> yeah, and its a pity that fedora doesnt know how to contribute
2007-07-14T21:44:13  * xorAxAx would have contributed the code if it had been a bottleneck in his wiki sites
2007-07-14T21:45:03  <grzywacz> Maybe they don't know it's trivial. :)
2007-07-14T21:45:20  <xorAxAx> we discussed the solution already on irc
2007-07-14T21:45:25  <grzywacz> ah so
2007-07-14T21:45:31  <xorAxAx> i presented a design
2007-07-14T21:46:14  <vpv> they don't actually "prefer" migration, it's just one option. but I hope someone would take the time to do it and at the same time I'll say "not gonna be me" until September ;)
2007-07-14T21:46:36  <vpv> by "do it" I mean contributing code to fix the problem
2007-07-14T21:47:08  <vpv> xorAxAx: who did you discuss it with? mmcgrath?
2007-07-14T21:48:14  <xorAxAx> yes
2007-07-14T21:52:47  <vpv> someone actually seems interested in looking into it, http://fedoraproject.org/wiki/Infrastructure/RFR/wiki/Infrastructure/RFR/MoinProfile
2007-07-14T21:53:11  <vpv> (what a weird page address btw ;)
2007-07-14T22:01:07  <grzywacz> Look broken.
2007-07-14T22:02:15  <grzywacz> *looks
2007-07-14T22:11:46  <xorAxAx> grzywacz: +    print "deleted" (This message has been postponed.)
2007-07-14T22:12:02  <grzywacz> ...
2007-07-14T22:12:04  <grzywacz> Ok.
2007-07-14T22:12:10  <xorAxAx> :)
2007-07-14T22:33:41  <ThomasWaldmann> re
2007-07-14T23:18:32  <ThomasWaldmann> johill: http://hg.moinmo.in/moin/1.7/rev/629dea07759f why?
2007-07-14T23:19:54  <xorAxAx> ThomasWaldmann: we had the discussion above
2007-07-14T23:20:05  <xorAxAx> ThomasWaldmann: and he posed a few questions for you
2007-07-14T23:20:44  * ThomasWaldmann scrolls back
2007-07-14T23:24:08  <ThomasWaldmann> well, using either single or double quotes can help for quoting the opposite quote (there is no escaping of those yet)
2007-07-14T23:27:56  <ThomasWaldmann> johill: so what is the reason for this changeset? above i only read about the troubles caused by it, but not why you did it.
2007-07-14T23:29:26  <xorAxAx> ThomasWaldmann: the idea was how to specify links
2007-07-14T23:29:34  <xorAxAx> ThomasWaldmann: didnt he send you a mail about that?
2007-07-14T23:31:06  <ThomasWaldmann> i have seen his mail, but didnt review it in detail yet
2007-07-14T23:31:26  <ThomasWaldmann> but a changeset comment should not depend on comments in some mail
2007-07-14T23:31:46  <xorAxAx> well, thats just one of a few changesets
2007-07-14T23:36:59  <ThomasWaldmann> wtf am I getting stuff to merge manually after pulling/updating from moin/1.7?
2007-07-14T23:37:31  <xorAxAx> yes
2007-07-14T23:37:34  <xorAxAx> i had the same issue
2007-07-14T23:38:07  <xorAxAx> some moved events.py file
2007-07-14T23:39:45  <vpv> iirc the manual work is useless, since the file is getting removed anyway
2007-07-14T23:40:04  <xorAxAx> yes
2007-07-14T23:40:31  <ThomasWaldmann> events/messages.py it is
2007-07-14T23:40:43  <xorAxAx> maybe somebody should give some reproducible command lines and feed them to #mercurial
2007-07-14T23:40:47  <xorAxAx> grzywacz: can you do that?
2007-07-14T23:40:58  <xorAxAx> you need to clone some revisions and then pull changes and update
2007-07-14T23:41:04  <xorAxAx> and it should see the conflict
2007-07-14T23:41:23  <xorAxAx> grzywacz: in order to know if its a bug or a broken merge of yourself
2007-07-14T23:41:31  <ThomasWaldmann> ehrm, no, notification.py
2007-07-14T23:41:52  <xorAxAx> grzywacz: branchview will guide you
2007-07-14T23:43:27  * ThomasWaldmann makes fresh clone
2007-07-14T23:46:28  <vpv> oh how nice, now I'm in the situation where I need to handle erroneus man pages in the manimport script :/
2007-07-14T23:47:08  <xorAxAx> now you need to show your robustness strengths
2007-07-14T23:47:27  <dreimark> vpv: humans are not programs
2007-07-14T23:47:54  <xorAxAx> they are, just faulty one
2007-07-14T23:47:55  <xorAxAx> s
2007-07-14T23:48:57  <dreimark> hehe
2007-07-14T23:49:34  <vpv> luckily manlifter does a pretty good job, I just need to react to it's status signals
2007-07-14T23:53:15  <xorAxAx> hmm, i think a brain is more like rewirable logic