MoinMoin: MoinMoinChat/Logs/moin-dev/2007-06-29

2007-06-29T00:02:47  <neagulm> starshine, something that could be used to shown in the info action and that should show the spam(junk) status for a page
2007-06-29T00:03:52  <starshine> so is it similar to the "updated" "deleted" series of icons then? "junked!"
2007-06-29T00:04:05  <neagulm> starshine, dreimark It would be nice an icon for the "notice bar" shown by the classification subsystem. The notice bar is similar to the notice bar from IE and FF
2007-06-29T00:04:34  <starshine> is this an icon for *showing* status or an icon for *setting* the status?
2007-06-29T00:05:45  <neagulm> starshine, dreimark I'm thinking...
2007-06-29T00:06:22  <neagulm> starshine, dreimark *showing*
2007-06-29T00:06:26  <neagulm> starshine, :)
2007-06-29T00:06:48  <starshine> ok and is it little square (e.g. diff icon) or long wordy (e.g. updated)
2007-06-29T00:07:12  <CIA-20> moin: Marian Neagul marian@info.uvt.ro * 2319:401b1eae8f21 1.7-classify-mneagul/MoinMoin/ (3 files in 2 dirs): Fix wikidict related bug
2007-06-29T00:08:33  <neagulm> starshine, I think that  little square would be ok! dreimark ?
2007-06-29T00:08:41  <dreimark> starshine: neagulm: I have to leave, long day/night tomorrow but i guess you find a nice solution without me
2007-06-29T00:08:48  <dreimark> neagulm: starshine yep
2007-06-29T00:09:20  <neagulm> dreimark, ok
2007-06-29T00:09:25  <neagulm> dreimark, good night
2007-06-29T00:09:46  <dreimark> neagulm: can you give ThomasWaldmann a note about that bug fix
2007-06-29T00:10:03  <dreimark> good night all
2007-06-29T00:15:23  <starshine> neagulm: so this is an indicator icon for the RC page?
2007-06-29T00:16:41  <neagulm> starshine, I need icons for the RC page and icons (spam/ham) for the notice bar that could appear for any page
2007-06-29T00:17:38  <starshine> what size are your icons for the notice bar, also which colors are acceptable?
2007-06-29T00:17:53  <neagulm> starshine, The best thing is to use the same icon set for both situations.
2007-06-29T00:18:08  <starshine> right, that's why I'm down to size and color-depth
2007-06-29T00:19:07  <neagulm> starshine, please see http://gsoc.info.uvt.ro/gsocwiki/MoinWord for the notice bar
2007-06-29T00:24:15  <starshine> I don't see a notice bar.  I have FF 1.5 at this location
2007-06-29T00:24:31  <starshine> and might have js turned off, I don't remember.
2007-06-29T00:25:55  <starshine> js on. java on. popups block.
2007-06-29T00:27:31  <neagulm> http://www.info.uvt.ro/~neagul/misc/noticebar.png
2007-06-29T00:28:11  <neagulm> starshine, I have marked with "ICON" the possible location for the icon...
2007-06-29T00:28:13  <starshine> oh.  That's the message block. ok.
2007-06-29T00:28:44  <starshine> that puts it at the same size (and reasonably, in the same table as) the smilies.
2007-06-29T00:29:21  <neagulm> starshine, yes
2007-06-29T00:36:07  <neagulm> starshine, I need to leave... I will be back tomorrow. :)
2007-06-29T00:36:12  <neagulm> starshine, good night
2007-06-29T00:36:23  <neagulm> good night all
2007-06-29T01:54:18  <xorAxAx> gn neagulm
2007-06-29T01:54:28  <xorAxAx> (hmm, a bit late :))
2007-06-29T01:59:39  <zenhase> oh ... two o'clock already
2007-06-29T02:01:10  <xorAxAx> gn zenhase
2007-06-29T02:02:34  <zenhase> well not right now for me
2007-06-29T02:03:05  <zenhase> but i think i will go home now, so i get to bed before 3 o'clock
2007-06-29T02:04:03  <zenhase> our talk this afternoon helped a lot btw.
2007-06-29T02:04:26  <zenhase> before that my head was really spinning around
2007-06-29T08:22:01  * grzywacz off to see if there's an exam to write
2007-06-29T10:19:32  <CIA-20> moin: Ville-Pekka Vainio <vpivaini AT cs DOT helsinki DOT fi> * 2291:b1abea506e62 1.7-maninfo-vpv/MoinMoin/script/xmlrpc/manimport.py: Really buggy url testing code, but I wanted to commit after a day's work
2007-06-29T10:19:35  <CIA-20> moin: Ville-Pekka Vainio <vpivaini AT cs DOT helsinki DOT fi> * 2292:91598a3873ae 1.7-maninfo-vpv/MoinMoin/script/xmlrpc/manimport.py: Some improvements to upcoming source package download feature. Don't send things in function arguments that are already in self.
2007-06-29T12:15:26  <xorAxAx> indeed, why isnt moin using itools.catalog?
2007-06-29T12:15:35  <xorAxAx> it looks like another migration will need to take place :)
2007-06-29T13:51:00  <zenhase> moin
2007-06-29T14:07:50  <zenhase> hmm
2007-06-29T14:07:59  <zenhase> plain python search index?
2007-06-29T14:09:10  <xorAxAx> zenhase: no
2007-06-29T14:09:59  <zenhase> i did not look at the slides
2007-06-29T14:10:41  <zenhase> but you think about using it for wiki searching
2007-06-29T14:10:50  <xorAxAx> no, to replace xapwrap
2007-06-29T14:11:23  <xorAxAx> moin already supports full text search
2007-06-29T14:11:32  <zenhase> i know
2007-06-29T14:11:55  <zenhase> thought you meant to externalize this
2007-06-29T14:12:11  <zenhase> but then, i am still in the process of waking up
2007-06-29T14:12:41  * zenhase gets some coffee
2007-06-29T14:26:37  <xorAxAx> zenhase: the issue of xapwrap is that it is not very well maintained and that it has a twisted dependency in the svn HEAD revision
2007-06-29T14:31:56  <CIA-20> moin: Ville-Pekka Vainio <vpivaini AT cs DOT helsinki DOT fi> * 2293:56d99f0e21e1 1.7-maninfo-vpv/MoinMoin/script/xmlrpc/manimport.py: Some more progress on the downloader, makes a temporary directory where it downloads the files
2007-06-29T15:37:20  <dreimark> xorAxAx: johill: the current tests for my changes to the wiki parser passes all so I can add it to 1.6 ?
2007-06-29T15:38:09  <johill> I guess so, no idea what the process is really
2007-06-29T15:38:16  <dreimark> ThomasWaldmann: I was on the wrong channel :)
2007-06-29T15:38:54  <xorAxAx> dreimark: thats the only sane policy, yes :)
2007-06-29T15:40:12  <ThomasWaldmann> dreimark: do you mean it passes the disabled tests? :D
2007-06-29T15:41:19  <dreimark> ThomasWaldmann: those too ;)  and it passes the new tests I added
2007-06-29T15:42:19  <dreimark> ok then I'll add it tommorrow
2007-06-29T15:44:44  <ThomasWaldmann> i assume you tested it with a cleared cache on some complex pages?
2007-06-29T15:47:14  <dreimark> yeah, SyntaxReference and some others, but I am not sure if someone has a more complex page as mine, but we can try then the testwiki afterwards
2007-06-29T15:49:18  <dreimark> ThomasWaldmann: neagulm has fixed in his branch a wikidict bug, (I have no log from yesterday night currently) did you have seen that?
2007-06-29T15:49:56  <ThomasWaldmann> afaics, that was no wikidict bug, but a fix of his code (deriving from wikidicts=
2007-06-29T15:52:16  <dreimark> ah ok
2007-06-29T15:52:37  <dreimark> have to leave now, back tommorrow
2007-06-29T18:08:39  <starshine> he could try a page with tables where inside one of the tables is an included page with more tables.  and have styles on them
2007-06-29T18:18:23  <CIA-20> moin: Ville-Pekka Vainio <vpivaini AT cs DOT helsinki DOT fi> * 2294:c4f51c5449a2 1.7-maninfo-vpv/MoinMoin/script/xmlrpc/manimport.py: Download the source file discarding the diffs and find the man pages from the archive. The files need to be extracted next and then passed to make_wiki_page()
2007-06-29T19:32:19  <xorAxAx> lanius_: do you have any questions about code coverage?
2007-06-29T20:06:14  <lanius> xorAxAx: no, thx
2007-06-29T20:06:41  <xorAxAx> but i want to answer a question about it :)
2007-06-29T20:06:59  <lanius> xorAxAx: why isn't it included in cpython?
2007-06-29T20:07:10  <xorAxAx> because its incompatible to unittest :)
2007-06-29T20:07:50  <lanius> but unittest sucks
2007-06-29T20:08:24  <xorAxAx> there are a lot of tests that need unittest
2007-06-29T20:09:13  <lanius> well, maybe in python 3k
2007-06-29T20:09:23  <xorAxAx> hmm?
2007-06-29T20:09:30  <lanius> it's not too far, read a report few days ago
2007-06-29T20:09:48  <xorAxAx> python 2.x and python 3k have a lot of unittests
2007-06-29T20:09:58  <xorAxAx> running using the unittest lib
2007-06-29T20:10:21  <lanius> i mean maybe they convert for 3k
2007-06-29T20:10:55  <lanius> http://osnews.com/story.php/18139/Python-3000-Status-Report
2007-06-29T20:11:09  <xorAxAx> ah
2007-06-29T20:11:11  <xorAxAx> yes
2007-06-29T20:11:17  <xorAxAx> that discussion is currently running
2007-06-29T20:11:34  <xorAxAx> but of course the new solution has to be able to run the old tests without a large refactoring
2007-06-29T20:11:49  <lanius> well, but it's 3k so you can break some things
2007-06-29T20:12:06  <xorAxAx> this is about workload, not about breakage
2007-06-29T20:13:18  <xorAxAx> many of the python tests dont use unittest yet
2007-06-29T20:13:26  <xorAxAx> because they are older than 6 years
2007-06-29T20:13:41  <xorAxAx> they simply call something and compare the stdout
2007-06-29T20:13:57  <lanius> yeah, simple approach
2007-06-29T20:25:01  <ThomasWaldmann> moin
2007-06-29T20:43:33  <grzywacz> moin
2007-06-29T23:22:04  <lanius> should the behaviour that editing a page which is in underlay will be copied to normal data be kept?
2007-06-29T23:24:50  <lanius> ThomasWaldmann: xorAxAx: and should renameItem and copyItem be added to the storage layer?
2007-06-29T23:27:32  <xorAxAx> lanius: yes for the first question
2007-06-29T23:27:39  <xorAxAx> lanius: yes for rename
2007-06-29T23:27:49  <xorAxAx> and no for copy
2007-06-29T23:28:15  <xorAxAx> well, at least not in the backend
2007-06-29T23:28:21  <xorAxAx> but maybe into the item collection
2007-06-29T23:28:44  <lanius> why do you want to keep that behaviour
2007-06-29T23:29:12  <xorAxAx> how else would you be able to edit the frontpage?
2007-06-29T23:29:30  <xorAxAx> of course you can rethink about underlay semantics
2007-06-29T23:29:49  <xorAxAx> the idea of underlay is that there is always only one version
2007-06-29T23:31:09  <lanius> you just edit the version in the underlay
2007-06-29T23:31:38  <lanius> not the version but the page in the underlay
2007-06-29T23:31:41  <lanius> why not
2007-06-29T23:31:47  <xorAxAx> no, thats not the wanted semantics
2007-06-29T23:31:58  <xorAxAx> because underlay is there to ease the upgrade for the admin
2007-06-29T23:32:06  <xorAxAx> he can replace the underlay directory by the new version
2007-06-29T23:32:23  <xorAxAx> and if you edited it in underlay, all your changes would be lost
2007-06-29T23:32:51  <lanius> but you can think of any "underlay" structure with the new backend
2007-06-29T23:33:03  <lanius> stuff
2007-06-29T23:33:10  <xorAxAx> yes, but you need to somehow fulfill this requirement
2007-06-29T23:33:15  <xorAxAx> maybe with saner semantics
2007-06-29T23:33:23  <lanius> saner semantics?
2007-06-29T23:33:31  <xorAxAx> yes, the 1.5-state is a bit insane
2007-06-29T23:33:40  <lanius> that means
2007-06-29T23:33:48  <xorAxAx> that it is broken by design
2007-06-29T23:34:09  <xorAxAx> cf. the revert spam to underlay page in first revision problem
2007-06-29T23:34:12  <xorAxAx> you cannot revert
2007-06-29T23:34:25  <xorAxAx> because the underlay revision will be magically gone
2007-06-29T23:34:34  <xorAxAx> you hvae to delete
2007-06-29T23:34:47  <xorAxAx> and if you delete, others will be confused because in fact the page is only reverted
2007-06-29T23:34:48  <lanius> i don't understand
2007-06-29T23:34:53  <xorAxAx> what dont you understand?
2007-06-29T23:34:59  <lanius> this spam stuf
2007-06-29T23:35:21  <xorAxAx> i think i have already explained it 2 times to you in here
2007-06-29T23:35:26  <lanius> and what semantic i have to clean up
2007-06-29T23:35:28  <xorAxAx> but nevertheless, the third time:
2007-06-29T23:35:47  <xorAxAx> imagine the frontpage in the underlay storage, now somebody comes and edits it with spam
2007-06-29T23:35:59  <xorAxAx> seombody else wants to revert it but sees only one revision in the revision list
2007-06-29T23:36:22  <xorAxAx> then he realises that he has to delete it - after he has deleted it, there is still one revision (the one from underlay)
2007-06-29T23:36:37  <xorAxAx> now somebody else looks at RC and sees that somebody deleted the page
2007-06-29T23:36:47  <xorAxAx> he looks at the page and wonders why it is still there
2007-06-29T23:37:02  <lanius> how do you think that can be fixed
2007-06-29T23:37:03  <xorAxAx> thats the old weird underlay semantics
2007-06-29T23:37:10  <xorAxAx> no idea :)
2007-06-29T23:37:25  <lanius> throw the underlay idea away
2007-06-29T23:37:35  <lanius> it's broken by design
2007-06-29T23:37:45  <xorAxAx> yeah, but with some other solution that fulfills the requirements
2007-06-29T23:37:56  <lanius> the only requirement is the update path?
2007-06-29T23:38:04  <xorAxAx> hmm, and sharing pages in a farm
2007-06-29T23:38:20  <xorAxAx> so somebody can have a wiki that is used as an underlay for the others
2007-06-29T23:38:57  <lanius> if you upgrade a wiki you don't want to upgrade the changed pages, so just don't overwrite them
2007-06-29T23:39:14  <lanius> a script could just look if there is more than one revision
2007-06-29T23:39:25  <lanius> and don't overwrite then
2007-06-29T23:39:30  <lanius> or just overwrite the first revision
2007-06-29T23:39:31  <xorAxAx> but there is no script based upgrade system for moin
2007-06-29T23:39:36  <xorAxAx> its all done manually currently
2007-06-29T23:39:57  <xorAxAx> yeah, you could overwrite the first revision
2007-06-29T23:40:28  <xorAxAx> but in fact, there will need to be scripts
2007-06-29T23:40:42  <lanius> that there currently is no script is not an argument
2007-06-29T23:40:46  <xorAxAx> otherwise the storage backend is useless (imagine somebody patching 600 files into an oracle db :))
2007-06-29T23:40:50  <lanius> to support a broken design
2007-06-29T23:40:52  <xorAxAx> yes, right
2007-06-29T23:41:06  <lanius> why do you want to update underlay at all
2007-06-29T23:41:18  <xorAxAx> because the help pages get updates
2007-06-29T23:42:14  <lanius> ok, use a script
2007-06-29T23:42:22  <xorAxAx> yes
2007-06-29T23:42:31  <xorAxAx> we already have "a script" :)
2007-06-29T23:42:36  <xorAxAx> the package management system
2007-06-29T23:42:45  <lanius> for databases you have to use a script anyway
2007-06-29T23:43:13  <lanius> and wiki farms could adapt the same behaviour
2007-06-29T23:43:18  <lanius> edit the page of the master wiki
2007-06-29T23:43:43  <xorAxAx> well, they could still use your stackable backend
2007-06-29T23:44:18  <lanius> they have to to get the pages of the master wiki
2007-06-29T23:44:27  <xorAxAx> hmm?
2007-06-29T23:44:41  <lanius> underlay is stackable backend as well
2007-06-29T23:45:40  <xorAxAx> with weird semantics above it
2007-06-29T23:45:57  <lanius> that's why i want to drop that
2007-06-29T23:46:11  <xorAxAx> well, the stackable backend could have two modes - edits go to the layer the stuff comes from and edits go the "uppest" layer
2007-06-29T23:46:18  <xorAxAx> i think the first one is still important
2007-06-29T23:46:31  <lanius> i only want to support the first one
2007-06-29T23:46:32  <xorAxAx> lanius: well, to have some manual copying for the second use case sucks
2007-06-29T23:46:47  <xorAxAx> yes, so the farm use case is fine with the second one IMHO
2007-06-29T23:46:59  <lanius> the second one is the underlay use case again
2007-06-29T23:47:04  <lanius> why not the first as well
2007-06-29T23:47:46  <xorAxAx> umm, i meant to say "first"
2007-06-29T23:47:50  <xorAxAx> sorry for confusion
2007-06-29T23:48:00  <lanius> k
2007-06-29T23:48:19  <lanius> so we agree to drop the copyUnderlay stuff
2007-06-29T23:48:26  <xorAxAx> yes
2007-06-29T23:48:29  <lanius> nice
2007-06-29T23:48:43  <xorAxAx> and components must accept that the first rev might have a date younger than the second rev
2007-06-29T23:48:55  <lanius> i see no problem there
2007-06-29T23:51:10  <lanius> why do you think copyPage should only be in the ItemCollection and not in the storage layer
2007-06-29T23:52:37  <lanius> it can be realized more performant on the storage layer
2007-06-29T23:52:47  <xorAxAx> thats the pythonic idea - python provides copying via a pure python external lib. for us the advantage is that we need copying support in the upper layer anyway as well - and what would we gain if it was in the lower layer?
2007-06-29T23:53:10  <xorAxAx> i dont think so
2007-06-29T23:53:26  <lanius> python copying objects doesn't help here
2007-06-29T23:53:33  <xorAxAx> yes, just a comparison
2007-06-29T23:53:38  <lanius> since you have to save the copy to the disk
2007-06-29T23:53:39  <xorAxAx> i dont think that this will slow it down
2007-06-29T23:53:44  <xorAxAx> give an example
2007-06-29T23:53:46  <lanius> which involves creating a new object from the ground up
2007-06-29T23:54:01  <xorAxAx> its much easier to implement it on upper levels
2007-06-29T23:54:11  <lanius> a copy would begin with ItemCollection.new_revision() (disk access)
2007-06-29T23:54:24  <lanius> then new_item i mean
2007-06-29T23:54:34  <lanius> then you do Item.new_revision() for each revision
2007-06-29T23:54:34  <xorAxAx> well, thats still nothing big
2007-06-29T23:54:41  <lanius> Item.revision[].setMetadata for each revision
2007-06-29T23:54:47  <lanius> item.revision[].writeData for each revision
2007-06-29T23:54:53  <lanius> and both must be read before
2007-06-29T23:54:57  <xorAxAx> yes, the backend has to do the same
2007-06-29T23:55:03  <xorAxAx> thats how copying works
2007-06-29T23:55:08  <lanius> the backend can do a filesytem copy operation
2007-06-29T23:55:18  <xorAxAx> which is the same io-wise and we are io-bound
2007-06-29T23:55:31  <xorAxAx> (even if optimised via fs-caches)
2007-06-29T23:56:17  <xorAxAx> you only get your copying op if you can proof that there is a sensible speed advantage by comparing both layers on a wiki page detailedly :)
2007-06-29T23:56:23  <lanius> i think opening a file, reading the content, opening another file, writing the content is not as performant as doing a copy
2007-06-29T23:56:37  <xorAxAx> thats wrong :)
2007-06-29T23:56:45  <lanius> i'll try
2007-06-29T23:56:46  <xorAxAx> posix doesnt have a copying op
2007-06-29T23:57:06  <xorAxAx> cp does open(), open(), read(), write(), read(), write(), ...
2007-06-29T23:57:16  <xorAxAx> and windows does the same, only behind a bunch of APIs on various levels
2007-06-29T23:57:32  <xorAxAx> should i look into the windows 2000 sources for you? :)
2007-06-29T23:57:57  <lanius> well i just can do some performance tests
2007-06-29T23:58:09  <xorAxAx> yes, but not now
2007-06-29T23:58:18  <xorAxAx> defer it and then you can still add it later on
2007-06-29T23:58:27  <xorAxAx> permature optimisation is the root of all evil
2007-06-29T23:58:32  <lanius> why
2007-06-29T23:58:36  <xorAxAx> and your only point is optimisation currently
2007-06-29T23:58:40  <lanius> right
2007-06-29T23:58:57  <xorAxAx> oh well, feed google with "permature optimisation evil"