MoinMoin: MoinMoinChat/Logs/moin-dev/2008-05-10

2008-05-10T10:05:58  <ThomasWaldmann> moin
2008-05-10T10:32:31  <zenhase> moin
2008-05-10T10:34:08  * ThomasWaldmann updates i188mn
2008-05-10T10:34:13  <ThomasWaldmann> gna
2008-05-10T10:40:35  <ThomasWaldmann> de(Recovery Token) = ?
2008-05-10T11:09:52  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3571:35cd758720e8 1.7/MoinMoin/i18n/ (42 files): updated i18n
2008-05-10T11:38:46  <kikka_> moin
2008-05-10T12:37:43  <dreimark> moin
2008-05-10T12:39:49  <dreimark> ThomasWaldmann: Wiederherstellungs Schlüssel / Zürücksetzungs password
2008-05-10T12:42:39  <ThomasWaldmann> versteht das dann noch jemand? :)
2008-05-10T12:42:51  <dreimark> nee
2008-05-10T12:43:07  <dreimark> wir hatten ja schon diskussionen wie man es im engl. nennt
2008-05-10T12:43:30  <ThomasWaldmann> Hmm, Passwort-uecksetzungs-Schluessel?
2008-05-10T12:43:35  <ThomasWaldmann> +R
2008-05-10T12:43:45  <dreimark> ich glaub im dt. kriegt man das nur mit einem satz hin
2008-05-10T12:44:01  <ThomasWaldmann> kommt halt recht oft vor in den neuen Texten
2008-05-10T12:44:03  <dreimark> aber das klingt schon ganz gut
2008-05-10T12:49:05  * ThomasWaldmann refactored MoinDev/Translation
2008-05-10T13:46:52  <ThomasWaldmann> dreimark: for the drawing test please use a browser/java that worked before that changeset
2008-05-10T13:58:48  * dreimark tries to reinstall/configure ff currently (damn 64bit)
2008-05-10T13:58:56  <dreimark> bbl
2008-05-10T15:26:47  <dreimark> ThomasWaldmann: verified browser/java combination with an old 1.6 that it has worked before
2008-05-10T15:27:00  <dreimark> now I do see the same string exception
2008-05-10T15:28:42  <dreimark> for 1.7 as already documented on the page
2008-05-10T15:32:41  * dreimark starts to debug it
2008-05-10T15:48:15  <dreimark> http://paste.pocoo.org/show/49145/
2008-05-10T15:48:37  <dreimark> lot's of Attribute errors by 'cStringIO.StringO'
2008-05-10T16:02:41  <dreimark> bbl
2008-05-10T16:18:03  * dennda tackles googles tax things and proof of enrollment
2008-05-10T16:29:22  <ThomasWaldmann> re
2008-05-10T16:36:01  <ThomasWaldmann> dennda: tax things? is it more than 10 words?
2008-05-10T16:38:45  * ThomasWaldmann looks at drawing stuff
2008-05-10T17:02:23  <dennda> ThomasWaldmann: The process involves cleaning my printers printing heads, printing, filling out, scanning and sending it as fax
2008-05-10T17:02:29  <dennda> (without having a fax machine)
2008-05-10T17:05:17  <ThomasWaldmann> :)
2008-05-10T17:10:53  <ThomasWaldmann> crap, the drawing stuff does not submit a filename when posting
2008-05-10T17:28:51  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3572:870cc4c47705 1.7/MoinMoin/action/AttachFile.py: AttachFile: fix saving of drawing files
2008-05-10T17:36:38  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2647:92729b7833eb 1.6/MoinMoin/action/AttachFile.py: AttachFile: fix saving of drawing files
2008-05-10T17:50:33  * ThomasWaldmann .oO(works for me now with opera9 and sun java 6)
2008-05-10T18:01:00  * dreimark works for my configuration too
2008-05-10T18:03:32  * ThomasWaldmann looks at the Category stuff now (or rather at improving those page_*_regexes)
2008-05-10T18:19:35  <dennda> boy, THAT was hard
2008-05-10T18:19:38  <dennda> just hit SENT
2008-05-10T18:24:08  <dennda> Your fax to Attn: Tiffany Griffith has been sent successfully
2008-05-10T18:24:11  <dennda> thanks, god
2008-05-10T18:25:35  <johill> fax? isn't that one of these ancient dead-tree product based technologise?
2008-05-10T18:34:10  * dreimark plays with xapian
2008-05-10T18:38:09  <dreimark> arrrgh I thought I have fixed that already AttributeError: 'NoneType' object has no attribute 'page_name'
2008-05-10T18:45:07  <dreimark> often i do see tracebacks of ShowSmileys.py, StatsChart.py
2008-05-10T18:51:44  <johill> we need something like kerneloops.org ;)
2008-05-10T18:52:01  <dennda> johill: yes, exactly what google asks you to use (and not use teh damn email)
2008-05-10T18:52:30  <johill> wanna bet they have a computer on the other end and no regular fax machine? ;)
2008-05-10T18:53:28  <dennda> it's just in order not to pollute LH's inbox
2008-05-10T18:53:39  <johill> heh
2008-05-10T18:53:40  <dennda> (how hard can it be to create a seperate alias...)
2008-05-10T18:54:05  <dennda> I used a free service on the net that adds ads somewhere on your page
2008-05-10T18:55:09  <ThomasWaldmann> that regex change will get a bit bigger
2008-05-10T18:56:04  <ThomasWaldmann> for example, I am thinking about change page_category_regex to ur'(Category(\w+))' as default
2008-05-10T18:56:56  <johill> what is it now?
2008-05-10T18:57:01  <ThomasWaldmann> so it will not only match category pages, but also find category page links within full text
2008-05-10T18:57:08  <johill> (and can't we do categories as real metadata soon?)
2008-05-10T18:57:30  <ThomasWaldmann> before it was '^Category[A-Z]'
2008-05-10T18:57:54  <ThomasWaldmann> johill: i am talking about 1.7 :)
2008-05-10T18:57:56  <johill> oh heh
2008-05-10T18:58:06  <johill> yeah still let's keep that in mind :)
2008-05-10T18:58:43  <ThomasWaldmann> strange. there is re.search, re.match, but no re.exactmatch?
2008-05-10T18:59:15  <johill> what would the difference between match and exactmatch be?
2008-05-10T18:59:55  <ThomasWaldmann> re.match(category_regex, 'CategoryFoo/WhatEver') matches
2008-05-10T19:00:33  <johill> and?
2008-05-10T19:00:49  <ThomasWaldmann> exactmatch matches not, because /WhatEver is not matched
2008-05-10T19:00:49  <johill> what do you expect from exactmatch?
2008-05-10T19:01:03  <johill> in '^Category[A-Z]'?
2008-05-10T19:01:04  <johill> why not?
2008-05-10T19:01:14  <ThomasWaldmann> no, with the new regex
2008-05-10T19:01:22  <johill> you're thinking of '^Category[A-Z][A-Za-z]*$'?
2008-05-10T19:01:22  <ThomasWaldmann> (18:56) < ThomasWal> for example, I am thinking about change page_category_regex to ur'(Category(\w+))' as default
2008-05-10T19:01:34  <johill> right
2008-05-10T19:01:57  <johill> but I don't see how any sort of regex function could guess what you need here?
2008-05-10T19:02:23  <johill> I mean, ^Category[A-Z] stops matching at CategoryF for CategoryFoo/whatever
2008-05-10T19:02:38  <ThomasWaldmann> if we search for category pages in the page list, it should only match on those that match exactly
2008-05-10T19:02:55  <ThomasWaldmann> (in old code, it uses search mostly)
2008-05-10T19:03:11  <johill> then you have to change the RE to what you proposed or maybe Category[A-Z]\w*
2008-05-10T19:03:16  <ThomasWaldmann> and it doesn't extract the category string by the regex
2008-05-10T19:04:28  <ThomasWaldmann> hmm, oops. i see that we had subcategories kind of. rather unintendedly.
2008-05-10T19:04:49  <johill> heh
2008-05-10T19:04:49  <ThomasWaldmann> bug or feature? :)
2008-05-10T19:05:04  <ThomasWaldmann> same for group
2008-05-10T19:05:07  <johill> no opinion really
2008-05-10T19:05:13  <ThomasWaldmann> FooPage/FooGroup
2008-05-10T19:05:26  <johill> group is a bit more interesting because of hierarchical ACLs
2008-05-10T19:05:40  <johill> that could get really confusing but also be a great feature
2008-05-10T19:06:01  <dreimark> I think we do use that for MyPages or ?
2008-05-10T19:06:10  <ThomasWaldmann> also, i don't want to have ^ and $ as part of the regex
2008-05-10T19:06:31  <ThomasWaldmann> (because then, it can't be used as is for search in full text)
2008-05-10T19:09:14  <dreimark> it should not become in conflict to the MyPages Action
2008-05-10T19:09:27  <ThomasWaldmann> dreimark: i didn't use that stuff recently. do we?
2008-05-10T19:09:47  <dreimark> I don't too I just read the page there
2008-05-10T19:10:00  <dreimark> it will use for example ReimarBauer/ReadWriteGroup
2008-05-10T19:11:59  <ThomasWaldmann> ah, right. ok, i need to tune the new regex
2008-05-10T19:16:14  <ThomasWaldmann> looks like I have to use \S+, not \w+
2008-05-10T19:17:26  <ThomasWaldmann> ok, i think that solves it, except for the ^ and $ stuff
2008-05-10T19:20:17  <ThomasWaldmann> e.g. not to match a page called "CategoryFoo is what I like" or "Suckers in EditorGroup"
2008-05-10T19:21:34  <ThomasWaldmann> -> back to exactmatch :)
2008-05-10T19:22:16  <johill> def _em(r):
2008-05-10T19:22:19  <ThomasWaldmann> a hack would be having 2 sets of precompiled regexes
2008-05-10T19:22:28  <johill>   return re.compile('^' + r + '$')
2008-05-10T19:22:32  <johill> re.exact_compile = _em
2008-05-10T19:22:34  <johill> ;)
2008-05-10T19:22:56  <johill> dinner, bbl
2008-05-10T19:23:16  <ThomasWaldmann> yeah, but if you dont want to compile that again and again, you have 2 sets of re objects in cfg.cache.foo
2008-05-10T19:24:58  <johill> can't you use re.search() and result.group(0)?
2008-05-10T19:25:04  <johill> to check if it was exact
2008-05-10T19:25:07  <johill> well, really bbl
2008-05-10T19:30:21  <ThomasWaldmann> johill: yes, but then we need a wrapper around it and can't just use regex.search as filter function as now
2008-05-10T19:31:14  <ThomasWaldmann> where's niemeyer when we need him :P
2008-05-10T23:26:48  <dreimark> ThomasWaldmann: ping
2008-05-10T23:27:03  <ThomasWaldmann> pong
2008-05-10T23:27:19  <dreimark> I think you can ask now
2008-05-10T23:27:36  <ThomasWaldmann> btw, i have failures with your new tests sometimes
2008-05-10T23:28:17  <ThomasWaldmann> i think the tests are dependant on timing
2008-05-10T23:29:45  <ThomasWaldmann> niemeyer: hi :)
2008-05-10T23:29:56  <niemeyer> ThomasWaldmann: Hey, good evening!
2008-05-10T23:30:28  <ThomasWaldmann> i had a regex problem :)
2008-05-10T23:31:19  <ThomasWaldmann> do you know if there is a special reason that re.search and re.match exist, but re.matchend and re.matchall don't?
2008-05-10T23:32:54  <niemeyer> ThomasWaldmann: Hmm
2008-05-10T23:33:40  <niemeyer> ThomasWaldmann: I guess matchall would be similar to findall()?
2008-05-10T23:33:53  <ThomasWaldmann> no
2008-05-10T23:34:03  <ThomasWaldmann> re.search matches somewhere in a string
2008-05-10T23:34:13  <ThomasWaldmann> re.match matches at the beginning only
2008-05-10T23:34:20  <ThomasWaldmann> re.matchend matches at end only
2008-05-10T23:34:42  <ThomasWaldmann> re.matchall only matches if there is nothing left (exact match, at beginning AND end)
2008-05-10T23:35:23  <niemeyer> ThomasWaldmann: Is there any reason for not using the regex markup for that?
2008-05-10T23:35:24  <ThomasWaldmann> (without that, you need ugly code as you will see in my next commit :)
2008-05-10T23:35:46  <ThomasWaldmann> yes, because I want to compile only once
2008-05-10T23:36:06  <niemeyer> ThomasWaldmann: I don't understand the relation
2008-05-10T23:37:06  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3573:124d0ef138aa 1.7/ (15 files in 10 dirs): change page_*_regex processing, see docs/CHANGES (fixes Xapian category search for non-english)
2008-05-10T23:37:21  <ThomasWaldmann> niemeyer: http://hg.moinmo.in/moin/1.7/rev/124d0ef138aa see there
2008-05-10T23:37:45  <ThomasWaldmann> read the docs/CHANGES hunk for details
2008-05-10T23:37:53  <niemeyer> ThomasWaldmann: In my personal view I see little reason even for search and match, as far as API goes
2008-05-10T23:39:21  <ThomasWaldmann> i had to store a complete second set of compiled regex objects.
2008-05-10T23:39:38  <ThomasWaldmann> just to have it do a "exact match".
2008-05-10T23:40:01  <ThomasWaldmann> as well as being able to find the same thing within some page text.
2008-05-10T23:43:34  <zenhase> re
2008-05-10T23:43:45  <niemeyer> ThomasWaldmann: Yeah, I understand the issue.. it's the price some times paid for having concise APIs
2008-05-10T23:45:08  <ThomasWaldmann> i just wondered about the assymetry.
2008-05-10T23:45:41  <ThomasWaldmann> match covers the match-at-beginning case, but there is not the opposite, and not both combined.
2008-05-10T23:46:17  <ThomasWaldmann> is there a better solution with the current api?
2008-05-10T23:48:05  <niemeyer> ThomasWaldmann: I have very vague memories about search() having an optimized algorithm for looking constant strings in the beginning of the string, so that it can very quickly get to the point where a candidate match starts
2008-05-10T23:48:21  <niemeyer> ThomasWaldmann: I might be wrong, but this might be the reason why we have two different functions
2008-05-10T23:48:59  <niemeyer> ThomasWaldmann: Arguably, this could be done by analysis, but I feel that this is the motivation for some reason
2008-05-10T23:51:17  <ThomasWaldmann> ok, so we do it like in that changeset. just though there could be maybe 2 easy member functions added to the regex objects to make such stuff easier.
2008-05-10T23:52:00  <ThomasWaldmann> s/though/thought/
2008-05-10T23:52:10  <niemeyer> ThomasWaldmann: Yeah, I understand your desire.. this is actually one of the reasons why I think there shouldn't be two functions, but rather one function that is able to analyze the initial portion of the regex and do the optimization whenever possible
2008-05-10T23:52:24  <niemeyer> ThomasWaldmann: If there was a single function, you'd probably not be looking for the others ;)
2008-05-10T23:53:27  <ThomasWaldmann> yeah :) if there was only search, it would be more consistent.
2008-05-10T23:55:13  <ThomasWaldmann> btw, str has startswith and endwith and == :P
2008-05-10T23:57:47  <ThomasWaldmann> dreimark: i guess the tests fails because it expects updated stuff a microsecond later, while some code might use (1s granularity) file timestamps (on edit-log)