MoinMoin: MoinMoinChat/Logs/moin-dev/2007-10-23

2007-10-23T00:38:22  <CIA-27> moin: Reimar Bauer <rb.proj AT googlemail DOT com> default * 2897:7f07db1306d1 1.7/MoinMoin/ (5 files in 4 dirs): makes ItemCache happy if shutil.rmtree is used
2007-10-23T00:39:28  <dreimark> wish to hear next time I am here that you get less failures in test_getVal. I hope that change fixes it for you.
2007-10-23T00:39:39  <dreimark> gn
2007-10-23T00:55:17  <grzywacz> night
2007-10-23T08:23:57  <ThomasWaldmann> moin
2007-10-23T11:32:02  <francesco_> hi, yesterday i asked if exists a way to change the normal edit function workflow. you answer me, that i1.6 implements textChas mechanism.
2007-10-23T11:33:49  <francesco_> the problem is: that now i'm implementing a validation mechanism for wiki and templating control, like an extension
2007-10-23T11:34:27  <francesco_> but now i can't control the normal edit workflow.
2007-10-23T11:34:41  <ThomasWaldmann> there is no code for extending this in 1.5. of course you could do a hack by backporting the 1.6 stuff.
2007-10-23T11:34:44  <francesco_> the textchas future textChas implementation
2007-10-23T11:35:00  <francesco_> provide stuff like hooks
2007-10-23T11:35:27  <francesco_> ?
2007-10-23T11:35:33  <francesco_> sorry...
2007-10-23T11:36:00  <ThomasWaldmann> (the editor code is rather ugly)
2007-10-23T11:36:01  <francesco_> the future textCHAS implementation will provide stuff like hooks,
2007-10-23T11:36:18  <ThomasWaldmann> there are no hooks yet
2007-10-23T11:36:24  <francesco_> umh...now im developing my graduation thesis
2007-10-23T11:37:03  <francesco_> and is it impossible to implement it, for research project?
2007-10-23T11:37:36  <francesco_> ahh Thomas... are you the creator of moinmoin wiki?
2007-10-23T11:40:41  <ThomasWaldmann> the creator was Juergen Hermann, but he is not active any more. I am the current project admin.
2007-10-23T11:41:01  <francesco_> ThomasWaldmann: where can i find 1.6 stuff to backporting in my 1.5?
2007-10-23T11:41:14  <ThomasWaldmann> hg.moinmo.in has all the code
2007-10-23T11:41:50  <ThomasWaldmann> but imho the editor could should get some cleanup before any major modification
2007-10-23T11:42:09  <ThomasWaldmann> see that textchas stuff more as a emergency hack
2007-10-23T11:42:36  <ThomasWaldmann> s/could/code/
2007-10-23T11:43:00  <francesco_> thank you, Thomas.
2007-10-23T11:43:16  <francesco_> Have you ever listen Light Constraint Templating
2007-10-23T11:43:18  <francesco_> ?
2007-10-23T11:43:43  <francesco_> Stefano Zacchiroli or Univeristy of Bologna, CS ?
2007-10-23T11:45:43  <TheSheep> urgh, it's about mediawiki "templates"
2007-10-23T11:46:11  <xorAxAx> ugh
2007-10-23T11:47:41  <francesco_> TheSheep: ?
2007-10-23T11:48:35  <francesco_> TheSheep: LC Templating mechanism is 'implementable' for every kind of wiki-engine, i hope.
2007-10-23T11:48:43  <blinx> moin
2007-10-23T11:48:49  <johill> dreimark: not sure I understand that patch
2007-10-23T11:51:47  <TheSheep> francesco_: excellent, we need more ways to constrain the wiki
2007-10-23T11:52:01  <blinx> ThomasWaldmann: ping
2007-10-23T11:53:39  <blinx> dreimark: any news about my bug =
2007-10-23T11:53:40  <blinx> ?
2007-10-23T11:55:17  <ThomasWaldmann> blinx: see latest 1.6 changesets. at least logging should work then.
2007-10-23T11:55:21  <francesco_> TheSheep: currently im working on Moin and MediaWiki, but Moin has not all the features of MediaWiki, and MediaWiki has not the modularity and flexibility of Moin.
2007-10-23T11:55:34  <blinx> searching for changes
2007-10-23T11:55:34  <blinx> no changes found
2007-10-23T11:55:48  <francesco_> TheSheep: it's a perfect situation
2007-10-23T11:56:49  <ThomasWaldmann> blinx: changeset af3caa3fe7ce
2007-10-23T11:57:18  <blinx> http://hg.moinmo.in/moin/1.6/rev/af3caa3fe7ce
2007-10-23T11:57:27  <xorAxAx> blinx: thats a question for #moin
2007-10-23T11:59:35  <blinx> pull is the keyword
2007-10-23T12:18:31  <francesco_> TheSheep: it is not a Costraint...
2007-10-23T12:19:32  <francesco_> it is LightConstraint, if you want to respect the template you can, but you can choose to not respect too
2007-10-23T12:19:38  <francesco_> to*
2007-10-23T12:20:50  <francesco_> is a trasparent way to help the users to respect a validation template...it's not properly a constraint, you can choose.
2007-10-23T12:23:23  <francesco_> ThomasWaldmann: I have implemented, Templating Validation as plugin/action/Validate.py , is there a way to link directly editform to this action? I need to perform validation after the first save, and add a step to the save action, and then report the validation results on the edit form ...
2007-10-23T12:24:36  <francesco_> i need stuff like hooks or similar to implement this!
2007-10-23T12:24:38  <francesco_> ??
2007-10-23T12:24:45  <ThomasWaldmann> the only hook yet is SecurityPolicy but that won't be enough
2007-10-23T12:25:08  <francesco_> umh...
2007-10-23T12:26:10  <francesco_> SecurityPolicy, where are antispam function and textchas
2007-10-23T12:26:44  <ThomasWaldmann> and as i said, before making action/edit.py, PageEditor.py and PageGraphicalEditor even more complex as they are currently,
2007-10-23T12:26:52  <francesco_> yes...but i don't want to change the source code to attach my validation on the edit funnction
2007-10-23T12:27:43  <ThomasWaldmann> there should be some refactoring, creating a universally usable editor widget class hierarchy
2007-10-23T12:28:40  <ThomasWaldmann> so that we have a simple text editor (mimetype text/*), a wiki text editor (text/moin-wiki), a gui wiki text editor (same) afterwards
2007-10-23T12:29:08  <ThomasWaldmann> (ah, forgot the rest editor (text/rst?))
2007-10-23T12:29:18  <francesco_> i understand...
2007-10-23T12:29:59  <ThomasWaldmann> and we have still storage refactoring code in that 1.7 branch that likely does some changes to PageEditor
2007-10-23T12:30:21  <ThomasWaldmann> so doing that before merging this would increase the workload for the merge
2007-10-23T12:30:31  <francesco_> ok...
2007-10-23T12:30:46  <xorAxAx> 1.7 has an event based system that allows you to write hooks, francesco_
2007-10-23T12:31:05  <ThomasWaldmann> but not interactively
2007-10-23T12:31:05  <francesco_> ok...
2007-10-23T12:31:10  <francesco_> yes...
2007-10-23T12:32:54  <francesco_> ok...the version 1.7 is currently available with this implementation , xorAxAx i can help us to realize it.
2007-10-23T12:34:55  <ThomasWaldmann> what exactly did you mean by your last sentence?
2007-10-23T12:39:14  <francesco_> for this university research project, to realize a complete validation system (like plugin, without change source) i need one hook into edit workflow, i have just finished to implement a validate action, and to complete the project i've need this hook.
2007-10-23T12:40:45  <francesco_> if this mechanism, event base system, is currently implemented on 1.7, i can help us to fixbug and add some hooks where will be useful
2007-10-23T12:41:36  <francesco_> i want only be useful...for me and for you
2007-10-23T12:42:40  <francesco_> ok, ThomasWaldmann, you understand?
2007-10-23T12:48:11  <ThomasWaldmann> yeah. the problem is order of things that need to get done (and whether you want to touch many other things except those you wanted to do for your research implementation)
2007-10-23T12:51:03  <francesco_> oh yes...
2007-10-23T12:54:12  <ThomasWaldmann> ok, so you could look at the amount of difference between Page*Editor in 1.7 and 1.7-storage
2007-10-23T12:54:49  <ThomasWaldmann> just to determine if we must merge that first to avoid lots of work
2007-10-23T12:55:54  <francesco_> i would like to give my graduation party in march, if i have time to spend for this work, but before i want to speak with my thesis relator
2007-10-23T12:57:52  <ThomasWaldmann> well, timing depends on the amount of work people are putting into the stuff that needs to get done
2007-10-23T12:58:32  <ThomasWaldmann> currently my personal main priority is to get 1.6 released until the end of the year
2007-10-23T12:59:15  <ThomasWaldmann> (but there are quite some other people that could work on other stuff)
2007-10-23T12:59:59  <francesco_> ThomasWaldmann, if i have enough time to spend for this project, and thesis works on the correct way, i would like to work for moin project
2007-10-23T13:00:38  <francesco_> ...for this and other stuff
2007-10-23T13:00:47  <ThomasWaldmann> any help is welcome :)
2007-10-23T13:01:23  <ThomasWaldmann> of course you need to communicate with us often, try to avoid decoupled work
2007-10-23T13:02:01  <francesco_> ok
2007-10-23T13:02:52  <francesco_> and ThomasWaldmann , last question for you about my personal work:
2007-10-23T13:04:03  <francesco_> in you opinion, you can exclude that exists a way to call an action in the editform by a plugin?
2007-10-23T13:04:55  <francesco_> ... a (n automatic) way to call an action in the editform by a plugin
2007-10-23T13:05:09  <ThomasWaldmann> there is no way yet to influence edit workflow in an interactive way
2007-10-23T13:05:54  <ThomasWaldmann> the only thing is either SecurityPolicy to avoid a save happening (e.g. for antispam or ACL restriction)
2007-10-23T13:06:14  <ThomasWaldmann> and 1.7 has events that get triggered AFTER the save happened
2007-10-23T13:06:59  <ThomasWaldmann> (you can just look at the file i quoted above to see that)
2007-10-23T13:07:08  <ThomasWaldmann> (you can just look at the files i quoted above to see that)
2007-10-23T13:11:19  <ThomasWaldmann> (btw, preview and spellcheck and switching editor type is an interactive part of the editing, but it is not a plugin, not even modular)
2007-10-23T13:14:06  <francesco_> i know, i know, i need modular way to attach my work to MoinMoin, like plugin. Ok... Thomas, thank you. This or next week i tell you if i can to start work for moin now or later, maybe with the beginning of next year? ok?
2007-10-23T13:14:47  <ThomasWaldmann> ok, just join us here :)
2007-10-23T13:18:57  <francesco_> ok :) thank you and sorry for had spend your time for my interests.
2007-10-23T13:20:21  <francesco_> sorry...having spent.
2007-10-23T13:58:13  <ThomasWaldmann> dreimark: getval test still fails when running on a fresh wiki. it does not fail when running the test on a non-fresh wiki.
2007-10-23T14:07:16  <dreimark> ThomasWaldmann: sounds like some progress
2007-10-23T14:07:42  <dreimark> how much failing tests did you have?
2007-10-23T14:07:59  <dreimark> johill: ping
2007-10-23T14:10:12  <CIA-31> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2233:8b1d72fbe560 1.6/MoinMoin/theme/__init__.py: navi_bar: highlight correct tab for the current page for i18ned system pages
2007-10-23T14:13:12  <CIA-31> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2234:d02aea9a8e26 1.6/MoinMoin/security/antispam.py: use logging module for antispam, not stderr
2007-10-23T14:13:12  *** irc.freenode.net sets mode: +o ChanServ
2007-10-23T14:18:08  <ThomasWaldmann> dreimark: 2/3
2007-10-23T14:18:48  <dreimark> before it was 4? and one is the _list[:] change by johill?
2007-10-23T14:24:32  <ThomasWaldmann> dreimark: now 4 are failing:
2007-10-23T14:24:51  <ThomasWaldmann> 2 because of not-changed page content when saving
2007-10-23T14:25:13  <ThomasWaldmann> 1 because of SystemPagesGroup == []
2007-10-23T14:25:30  <ThomasWaldmann> 1 is GetVal
2007-10-23T14:26:18  <dreimark> ThomasWaldmann: the last one was caused from changeset http://hg.moinmo.in/moin/1.7/rev/7e4627efd98e
2007-10-23T14:26:29  <dreimark> not the last one the  of SystemPagesGroup == []
2007-10-23T14:26:56  <dreimark> this one I have too.
2007-10-23T14:28:09  <ThomasWaldmann> johill: could you improve that?
2007-10-23T14:32:19  <dreimark> ThomasWaldmann: the tests should always be able to save because at the end of the test the pagedir is removed and a log entry should be done so ItemCache is clean too
2007-10-23T14:33:05  <dreimark> can you please list the pagedir.
2007-10-23T14:33:10  <dreimark> is it really there
2007-10-23T14:39:15  <CIA-31> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2235:37ae4901365f 1.6/MoinMoin/config/multiconfig.py: auto-disable xapian_search when xapian fails to import
2007-10-23T14:51:34  <ThomasWaldmann> dreimark: http://docs.moinmo.in/files/dreimark/
2007-10-23T14:55:23  <dreimark> reading
2007-10-23T15:35:05  <johill> dreimark: pong
2007-10-23T15:35:16  <dreimark> :)
2007-10-23T15:35:25  <johill> ThomasWaldmann: huh tests fail because I changed the order of something??
2007-10-23T15:35:48  <dreimark> no because you replaced keys by this
2007-10-23T15:35:52  <dreimark> _list thing
2007-10-23T15:36:02  <johill> hm but that should only change the order, no?
2007-10-23T15:36:29  <johill> I'll look at it, only have a minute now need to catch some people in a few minutes
2007-10-23T15:36:43  <dreimark> ok
2007-10-23T15:36:50  <johill> back in 10-15 minutes or so
2007-10-23T15:37:10  <dreimark> ok, same for me
2007-10-23T15:38:50  <dreimark> ThomasWaldmann: do you have time to to read 38 pages about MM coding in german the next days?
2007-10-23T15:39:30  <ThomasWaldmann> just send it to me, I'll try
2007-10-23T15:39:40  <dreimark> :)
2007-10-23T15:39:41  <ThomasWaldmann> or put it on some wiki page
2007-10-23T15:47:28  <CIA-31> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2236:3ece805ba035 1.6/MoinMoin/ (PageGraphicalEditor.py security/textcha.py): TextChas for GUI editor, update TODO
2007-10-23T15:51:42  <johill> ok, back
2007-10-23T15:52:11  * johill runs tests
2007-10-23T15:59:33  <johill> ok but surely I didn't cause all these regressions
2007-10-23T16:01:23  <ThomasWaldmann> not all, but maybe 1 :)
2007-10-23T16:01:52  <johill> yeah, I can see the re-reading dict after page goes away
2007-10-23T16:05:06  <francesco_> alla prossima
2007-10-23T16:09:22  <johill> grr
2007-10-23T16:09:25  <johill> who wrote these tests?
2007-10-23T16:09:27  <johill> now they just fail with
2007-10-23T16:09:36  <johill> You did not change the page content, not saved
2007-10-23T16:10:31  <xorAxAx> which one?
2007-10-23T16:11:50  <johill> MoinMoin/_tests/test_wikidicts.py
2007-10-23T16:11:56  <johill> both that do saveText()
2007-10-23T16:12:09  <johill> testRenameGroupPage and testCopyGroupPage
2007-10-23T16:12:28  <johill> and if I catch that they crap out every second time
2007-10-23T16:12:30  <johill> grrrr
2007-10-23T16:12:34  <johill> crap tests
2007-10-23T16:14:13  <johill> they were written to fail testing if the thing to test ever fails
2007-10-23T16:14:52  <xorAxAx> hehe
2007-10-23T16:15:07  <dreimark> johill: thats my tests and they did not fail on all three of my systems
2007-10-23T16:15:09  <xorAxAx> isnt the goal to reduce the number of failing tests? )
2007-10-23T16:15:20  <johill> dreimark: yeah, but if they ever fail the test breaks horribly
2007-10-23T16:15:31  <TheSheep> xorAxAx: empty files are correct programs ;)
2007-10-23T16:15:32  <johill> where it'll then fail every time with an unrelated failure
2007-10-23T16:15:44  <johill> or if I fix  that, fail every second time only
2007-10-23T16:15:53  <johill> iow, the tests don't clean up after themselves
2007-10-23T16:16:02  <johill> you need to fix that so they clean up even if they fail
2007-10-23T16:16:13  <dreimark> johill: at my system they did
2007-10-23T16:16:28  <johill> I don't really know what's wrong
2007-10-23T16:16:41  <johill> all I can tell you is that once they failed once as they currently do, they fail in page saving the second time around
2007-10-23T16:16:42  <dreimark> and yesterday I thought they did not clean up on yours because I have not killed the ItemCache right
2007-10-23T16:16:44  <johill> which is totally bogus
2007-10-23T16:16:50  <xorAxAx> TheSheep: exactly
2007-10-23T16:18:09  <dreimark> they clean up here everytime and that tells a story
2007-10-23T16:18:39  <johill> well, here they fail with "Unchanged: you didn't change the page"
2007-10-23T16:18:44  <dreimark> I dont get why it doesnt on your systems but always on mine which are very different
2007-10-23T16:18:47  <johill> and I didn't change anything over 1.7 yet
2007-10-23T16:19:15  <johill> and if I catch the Unchanged exception, they run ok, fail the second time, run ok again etc
2007-10-23T16:19:17  <dreimark> johill: the page dir is removed after the test, the itemcache is cleared
2007-10-23T16:19:39  <johill> no it's not removed after the test
2007-10-23T16:19:40  <johill> see
2007-10-23T16:19:42  <johill> you remoev it in the test
2007-10-23T16:19:47  <johill> but the test raises an exception before that
2007-10-23T16:19:48  <xorAxAx> dreimark: writing a wikidicts that does not fail is the first problem ...
2007-10-23T16:19:56  <xorAxAx> dreimark: you always need to have a failing test
2007-10-23T16:19:57  <johill> you're missing a try: finally: clause there for the cleanup
2007-10-23T16:20:02  <xorAxAx> otherwise its useless to check it in
2007-10-23T16:20:10  <xorAxAx> and wikidicts have known issues
2007-10-23T16:20:28  <johill> dreimark: see, you're cleaning up in the case that the 'assert' failes
2007-10-23T16:20:31  <xorAxAx> (which doesnt ease the process of getting a failing test necessarily)
2007-10-23T16:20:39  <johill> dreimark: but you're not cleaning up when there's a bug and teh test fails due to ean exception
2007-10-23T16:20:57  <johill> so your test assumes that the code runs fine but gives the wrong result
2007-10-23T16:21:01  <xorAxAx> dreimark: because if the test doesnt fail, you cannot tell whether it is correct
2007-10-23T16:21:01  <johill> rather than assuming that anything can go wrong
2007-10-23T16:23:16  <dreimark> I'll add the cleanup, but i dont get why I never all the time have had those errors you got
2007-10-23T16:23:44  <johill> no idea
2007-10-23T16:23:49  <johill> maybe you blew away the data dir every time?
2007-10-23T16:23:53  <dreimark> no
2007-10-23T16:24:38  <dreimark> at first time it creates the page and removes it
2007-10-23T16:24:50  <dreimark> that is repeated for the next one
2007-10-23T16:25:27  <johill> no see it doesn't remove it if there's an exception
2007-10-23T16:25:29  <TheSheep> these tests fail in a very strange and undeterministic way on my box
2007-10-23T16:25:32  <johill> so on the next time the page is still around
2007-10-23T16:25:37  <johill> TheSheep: yeah, same here
2007-10-23T16:25:52  <johill> it seems that somehow there are timing/locking issues
2007-10-23T16:26:04  <TheSheep> johill: try adding some more detailed asserts about names being in certain groups, it's madness
2007-10-23T16:26:35  <johill> the thing I don't understand is why my patch would break it
2007-10-23T16:26:48  <johill> these wikidicts are never re-initialised are they?
2007-10-23T16:26:52  <TheSheep> johill: sometimes the page is in the list but not in the dicts, sometimes in both and sometimes in none
2007-10-23T16:27:09  <TheSheep> johill: no, these tests fail since quite a while
2007-10-23T16:27:17  <dreimark> TheSheep: i tried to change that with the patch in the pastebin
2007-10-23T16:27:45  <dreimark> johill: there is one test of wikiutils which is now failing
2007-10-23T16:27:56  <johill> ah
2007-10-23T16:28:03  <TheSheep> dreimark: which pastebin?
2007-10-23T16:28:26  <johill> dreimark: that pastebin is obviously incorrect as it reverts my original patch
2007-10-23T16:28:29  <johill> at least partially
2007-10-23T16:28:35  <johill> since it again kills the ordering
2007-10-23T16:28:45  <johill> http://paste.pocoo.org/show/7109/
2007-10-23T16:28:48  <johill> -        return self._list[:]
2007-10-23T16:28:48  <johill> +        return self.keys()
2007-10-23T16:28:56  <dreimark> recognized this too
2007-10-23T16:28:56  <johill> the only *reason* I did the _list was to use it there
2007-10-23T16:29:06  <johill> (and in iter())
2007-10-23T16:29:22  <TheSheep> dreimark: ah, that's another one :)
2007-10-23T16:32:09  <dreimark> I do want to get http://paste.pocoo.org/show/7211/
2007-10-23T16:32:28  <dreimark> a list for a user
2007-10-23T16:33:04  <dreimark> and have not seen how it should work without my changes
2007-10-23T16:33:36  <johill> hrm
2007-10-23T16:33:39  <johill> somehow this is borked
2007-10-23T16:33:45  <johill> why does the systme groups dict not init from text?!
2007-10-23T16:36:43  <dreimark> btw. is there any graphically analysing tool for dicts available.
2007-10-23T16:37:15  <dreimark> I didn't get why it doesnt use _list from init
2007-10-23T16:41:55  <johill> ?
2007-10-23T16:43:30  <johill> ok I found the bug, it uses update()
2007-10-23T16:44:09  <dreimark> wow :)
2007-10-23T16:45:13  <johill> oddly, overriding update() doesn't help?!
2007-10-23T16:46:18  <johill> hrm
2007-10-23T16:46:22  <johill> no, that's just one part of the bug
2007-10-23T16:48:33  <johill> or it's caching stuff
2007-10-23T16:50:35  <johill> ah it is
2007-10-23T16:50:36  <johill> crap
2007-10-23T16:50:56  <johill> tests that rely on changes.. why does the cache kick in?
2007-10-23T16:51:30  <johill> again, a test that behaves erratically
2007-10-23T16:51:36  <johill> sometimes it prints HUGE lists and sometimes it just fails
2007-10-23T16:51:37  * ThomasWaldmann adds textcha to AttachFile
2007-10-23T16:51:39  <johill> depending on the cache
2007-10-23T16:51:42  <johill> hrm
2007-10-23T16:51:50  <johill> you could fix the cache instead ;)
2007-10-23T16:52:04  <xorAxAx> the itemcache? hehe
2007-10-23T16:52:11  <xorAxAx> you can try to run it wihtout it
2007-10-23T16:52:25  <dreimark> xorAxAx: do we have a config var to disable it?
2007-10-23T16:52:46  <xorAxAx> dreimark: no, but its only one line of code in the config
2007-10-23T16:52:49  <xorAxAx> hmm
2007-10-23T16:53:07  <dreimark> i would prefer a config var and a description on MMaster
2007-10-23T16:53:30  <xorAxAx> but i dont have it in one of my of wikis
2007-10-23T16:53:43  <xorAxAx> request.cfg.disable_50_percent_of_bugs
2007-10-23T16:53:59  <CIA-31> moin: Johannes Berg <johannes AT sipsolutions DOT net> default * 2898:dd2226c7f5fe 1.7/MoinMoin/wikidicts.py: fix wikidicts by overriding Group.update()
2007-10-23T16:53:59  <dreimark> hehe, negative image
2007-10-23T16:54:20  <johill> that fixes the test I made fail, I hope
2007-10-23T16:54:33  <johill> pretty hard to tell due to the erratic behaviour of  the test
2007-10-23T16:54:44  <johill> but it works with and without cache now
2007-10-23T16:54:48  <ThomasWaldmann> request.cfg.kill_the_harddisk :)
2007-10-23T16:55:16  <dreimark> yeah, thats the same feeling I have with those getVal and Dict tests I didnt get failing here
2007-10-23T16:55:37  <johill> possible
2007-10-23T16:55:54  <ThomasWaldmann> btw, I suggest changine that cache stuff AFTER we have merged new storage code
2007-10-23T16:55:55  <xorAxAx> dreimark: wrong file system
2007-10-23T16:56:06  <xorAxAx> ThomasWaldmann: in the new storage gone, it was killed
2007-10-23T16:56:12  <xorAxAx> s/gone/stuff/
2007-10-23T16:56:17  <xorAxAx> ugh, i should stop chatting
2007-10-23T16:56:41  <ThomasWaldmann> ok, so nothing to do then. except working on making the merge possible
2007-10-23T16:56:42  <dreimark> xorAxAx: I have XFS and EXT3 without problems
2007-10-23T16:57:08  <TheSheep> they still fail for me
2007-10-23T16:58:09  * ThomasWaldmann really looks forward to that new storage backend, but I can't do much work on it now.
2007-10-23T16:58:09  <dreimark> ThomasWaldmann: yesterday i recognized that its not possible to change pagedirs on filesystem level if itemcache is running always
2007-10-23T16:58:24  <dreimark> and you have cgi
2007-10-23T16:59:04  <ThomasWaldmann> it relies on edit-log for invalidation
2007-10-23T17:00:07  <johill> now MoinMoin/formatter/_tests/test_pickle.py fails
2007-10-23T17:00:10  <ThomasWaldmann> the work needed, btw, is solve attachments (== continue the new fs storage backend that i have begun)
2007-10-23T17:00:19  <ThomasWaldmann> and implement ACLs
2007-10-23T17:00:33  <johill> haha
2007-10-23T17:00:34  <johill> NameError: global name 'py' is not defined
2007-10-23T17:01:02  <dreimark> if you choose non persistent servers you may be dont want internally persistent
2007-10-23T17:01:18  <johill> oh wait, that's my code to start with
2007-10-23T17:01:29  <johill> that's just this pickle formatter stuff I did a long time ago
2007-10-23T17:01:41  <ThomasWaldmann> dreimark: why that?
2007-10-23T17:02:32  <dreimark> because itemcache behaves similiar to fcgid running all is done in memory and you will restart the server
2007-10-23T17:02:53  <dreimark> if you did changes on filesystem level
2007-10-23T17:03:22  <dreimark> which is currently not needed
2007-10-23T17:03:25  <ThomasWaldmann> well, there are 2 possibilities:
2007-10-23T17:03:46  <ThomasWaldmann> a) cache some stuff in-process and invalidate it correctly
2007-10-23T17:03:59  <ThomasWaldmann> b) don't do it and access the disk all the time
2007-10-23T17:04:33  <ThomasWaldmann> b is rather slow because even in a single request, the same stuff is accessed again and again
2007-10-23T17:04:50  <xorAxAx> the disk is not accessed ... but the fs cache
2007-10-23T17:04:56  <dreimark> but b) will not cost memory
2007-10-23T17:05:26  <dreimark> sometimes cgi machines dont have much memory
2007-10-23T17:05:29  <xorAxAx> (except for a cold cache)
2007-10-23T17:05:43  <ThomasWaldmann> xorAxAx: i did quite some experiments before implementing that itemcache
2007-10-23T17:06:12  <xorAxAx> well, might be the case :)
2007-10-23T17:06:30  <xorAxAx> did you try invalidating via the file size?
2007-10-23T17:07:01  <ThomasWaldmann> iirc, latest code checks size of edit-log, not mtime
2007-10-23T17:07:02  <xorAxAx> that should change on any fs, regardless how fast the update is done
2007-10-23T17:07:05  <xorAxAx> hmm
2007-10-23T17:07:32  <xorAxAx> (but it would still have nfs issues because of nfs caches)
2007-10-23T17:07:59  <ThomasWaldmann> that's why it is called network failure system :))
2007-10-23T17:09:13  <dreimark> how much time does it cost to build the itemcache?
2007-10-23T17:09:37  <ThomasWaldmann> it builds itself on demand
2007-10-23T17:10:16  <xorAxAx> updating it is O(n^2)
2007-10-23T17:10:28  <xorAxAx> in some cases
2007-10-23T17:10:40  <xorAxAx> because its using a for+del loop
2007-10-23T17:10:47  <dreimark> that timeout that blinx has had with a lot of [[ ]] which are now used for linking on his old underlay c
2007-10-23T17:11:17  <dreimark> could that be caused from the itemcache ?
2007-10-23T17:12:03  <ThomasWaldmann> the issues he had were mainly the i18n cache and not installing stuff correctly
2007-10-23T17:12:29  <dreimark> yeah I know, but do we know limits of itemcache
2007-10-23T17:12:41  <ThomasWaldmann> the i18n stuff still needs fixing, some other minor issues are already fixed (in 1.6)
2007-10-23T17:13:10  <ThomasWaldmann> itemcache will go anyway
2007-10-23T17:13:41  <ThomasWaldmann> (but not now, due to the todo i posted above)
2007-10-23T17:13:46  <dreimark> ok
2007-10-23T17:14:21  <dreimark> for i18n the easiest solution is to set formatted=False
2007-10-23T17:14:29  <ThomasWaldmann> no
2007-10-23T17:14:41  <ThomasWaldmann> (it doesnt help with that problem)
2007-10-23T17:15:02  <ThomasWaldmann> because the formatting happens when the texts are loaded (all at once)
2007-10-23T17:15:22  <ThomasWaldmann> not when you call getText with formatted=True
2007-10-23T17:16:29  <ThomasWaldmann> but you get me thinking ...
2007-10-23T17:16:32  <dreimark> which bug report ?
2007-10-23T17:16:55  <ThomasWaldmann> see the log output of 1.6 X)
2007-10-23T17:17:32  <ThomasWaldmann> it's completely insane, but the lock code I added made it even worse (thus I removed it again)
2007-10-23T17:18:16  <dreimark> http://moinmo.in/MoinMoinBugs/1.6ErrorWhenFindNoResults is similiar
2007-10-23T17:19:11  <dreimark> ThomasWaldmann: pdf in mail
2007-10-23T17:19:33  <ThomasWaldmann> dreimark: that's another i18n code issue
2007-10-23T17:20:08  <ThomasWaldmann> pdf is rather unsuitable for doing comments
2007-10-23T17:20:27  <dreimark> I can give you latex
2007-10-23T17:20:33  <dreimark> is that ok?
2007-10-23T17:20:54  <ThomasWaldmann> ehrm, yes, additionally
2007-10-23T17:23:24  <ThomasWaldmann> dreimark: btw, in some of your i18n texts, I think there are macro calls. They should be like `<<...>>` except if you want them executed.
2007-10-23T17:24:50  <ThomasWaldmann> 0824: "Not enough arguments given to EmbedObject macro! Try <<EmbedObject"
2007-10-23T17:25:13  <johill> unless you use formatted=False
2007-10-23T17:27:01  <ThomasWaldmann> no. they are formatted once on load.
2007-10-23T17:27:13  <johill> ?
2007-10-23T17:27:41  <ThomasWaldmann> the load does load all text from the .po
2007-10-23T17:27:41  <johill> if you have a string and pass it to _(, formatted=False) it shouldn't be formatted
2007-10-23T17:28:00  <ThomasWaldmann> then converts to .mo internally and makes some Translation object with it
2007-10-23T17:28:03  <johill> and formats it all?
2007-10-23T17:28:08  <johill> that's bogus
2007-10-23T17:28:11  <ThomasWaldmann> and then formats all of them
2007-10-23T17:28:27  <ThomasWaldmann> an the po has no information whether it is used formatted or not
2007-10-23T17:28:31  <johill> so formatted=False has become useless now
2007-10-23T17:28:37  <johill> except for the original english text
2007-10-23T17:28:37  <ThomasWaldmann> all the result is then cached to disk
2007-10-23T17:28:39  <johill> that's dumb
2007-10-23T17:28:50  <ThomasWaldmann> no, it is not useless
2007-10-23T17:29:01  <ThomasWaldmann> it determines which cache to use
2007-10-23T17:29:10  <johill> ??
2007-10-23T17:29:22  <johill> so you have one cache with and one without formatting?
2007-10-23T17:29:24  <ThomasWaldmann> the cache has unformatted and formatted text in it
2007-10-23T17:29:37  <johill> but then using no `` is correct
2007-10-23T17:30:13  <ThomasWaldmann> hmm, right
2007-10-23T17:30:26  <johill> I think formatting at cache time is wrong
2007-10-23T17:30:31  <ThomasWaldmann> well, we obviously need a better approach :)
2007-10-23T17:30:33  <johill> you have no user information
2007-10-23T17:31:05  <ThomasWaldmann> but if you format it on call time, it costs quite some performance
2007-10-23T17:31:46  <ThomasWaldmann> one could update the cache then
2007-10-23T17:33:01  <johill> no, because macros may have dependencies on users
2007-10-23T17:33:11  <johill> we should ideally have almost everything with formatted=False anyway
2007-10-23T17:33:43  <ThomasWaldmann> there is lots of <<BR>>
2007-10-23T17:34:26  <johill> that could be cahed
2007-10-23T17:34:28  <johill> cached
2007-10-23T17:34:36  <johill> but we don't have proper macro dependencies
2007-10-23T17:34:50  <johill> that goes back to the stuff I did with request.state
2007-10-23T17:34:56  <johill> s/did/thought of/
2007-10-23T17:35:18  <johill> we really need better state handling, then we can automatically have better dependency handling
2007-10-23T17:35:37  <ThomasWaldmann> yeah, but not for 1.6 :)
2007-10-23T17:36:03  <johill> true
2007-10-23T17:36:12  <johill> but then I guess we have to take the slower but correct way
2007-10-23T17:36:25  <johill> or make sure none of the macros use user state
2007-10-23T17:36:34  <johill> then we can cache on first formatting
2007-10-23T17:36:38  <ThomasWaldmann> i will just remove the formatting from the load call and do it on demand in getText and we'll see
2007-10-23T17:37:23  <ThomasWaldmann> as it looks, we'll have to implement a special formatter for that other bug anyway, so we can do some stuff there
2007-10-23T17:37:50  <johill> haven't really looked at the bug
2007-10-23T17:38:10  <ThomasWaldmann> links with nonascii stuff have %XX in them
2007-10-23T17:38:23  <ThomasWaldmann> if you use that as a left side of %, you have trouble
2007-10-23T17:38:49  <johill> makes sense
2007-10-23T17:38:54  <johill> but why would that ever happen?
2007-10-23T17:39:11  <johill> because of using formatted text that included a link?
2007-10-23T17:39:19  <dreimark> because of wikinames which are exapnded to Urls
2007-10-23T17:39:32  <dreimark> links
2007-10-23T17:39:43  <johill> ok so in that case you want formatted=False and placeholders?
2007-10-23T17:39:51  <johill> I mean formatted=True
2007-10-23T17:42:16  <johill> I suppose for that we'll have to introduce expanded=True or so, and if yes expand all % chars that aren't in the original text
2007-10-23T17:42:25  <johill> expand them to %% of course
2007-10-23T17:43:04  <ThomasWaldmann> or some special link formatting
2007-10-23T17:43:11  <ThomasWaldmann> s/or/and/
2007-10-23T17:44:16  <johill> why?
2007-10-23T17:44:30  <johill> actually, yeah, that's part of the "that aren't in the original text"
2007-10-23T17:45:06  <ThomasWaldmann> brb
2007-10-23T18:07:47  <dreimark> ThomasWaldmann: May be we should add a link to the Attachment dialog to HelpOnLinking or/and write the explaination differently
2007-10-23T18:09:50  <dreimark> To refer to attachments on a page, use [[attachment:filename]],
2007-10-23T18:10:11  <dreimark> currently we have To refer to attachments on a page, use attachment:filename,
2007-10-23T18:45:06  <dreimark> bbl
2007-10-23T18:55:10  <CIA-31> moin: Thomas Waldmann <tw AT waldmann-edv DOT de> default * 2237:51ce8385c960 1.6/ (MoinMoin/action/AttachFile.py docs/CHANGES): Add TextChas to attachment upload, add textcha to CHANGES, cleanup/reorder AttachFile handler
2007-10-23T18:56:24  <xorAxAx> ThomasWaldmann: "Once your payment is requested it can take up to 30 days to get your check. Google pays out on a net 30 so depending where they are in the cycle when the request was turned in will determine how long it takes to get your check. "
2007-10-23T18:56:29  <xorAxAx> they seem to need a req
2007-10-23T18:56:35  <xorAxAx> (for whatever reason?)
2007-10-23T18:58:56  <ThomasWaldmann> xorAxAx: is sending them a mail like I did not "a request"?
2007-10-23T19:00:19  <xorAxAx> ThomasWaldmann: i think they talk about the org payment which is something different than the travel payment
2007-10-23T19:00:43  <xorAxAx> (the travel stuff seems to have a 7 day interval and now she is talking about 30 days)
2007-10-23T19:11:58  <ThomasWaldmann> request sent
2007-10-23T20:26:59  <CIA-31> moin: Reimar Bauer <rb.proj AT googlemail DOT com> default * 2899:5a3037bd131a 1.7/MoinMoin/ (_tests/test_wikidicts.py macro/_tests/test_GetVal.py): test_wikidicts, test_GetVal: avoid error if pagecontent could not be saved
2007-10-23T20:49:41  <dreimark> ThomasWaldmann: we normally dont want one to use the back button in the browser, text cha gives a hint to do that
2007-10-23T20:53:30  <ThomasWaldmann> yeah
2007-10-23T20:54:06  <dreimark> currently text cha seems to be enabled for 1.6 anonymous edit
2007-10-23T20:54:09  <ThomasWaldmann> (as the save did not work in that case, it should work ok, for non-broken browsers)
2007-10-23T20:54:35  <ThomasWaldmann> it is disabled by default
2007-10-23T20:54:48  <ThomasWaldmann> and if you enable it, it is on for everybody
2007-10-23T20:55:37  <dreimark> its enabled in wikiconfig.py of the standalone server currently
2007-10-23T20:55:56  <ThomasWaldmann> yeah, that's for trying, see that comment
2007-10-23T20:55:57  <dreimark> but I read there for trying out
2007-10-23T20:57:29  <dreimark> have to register
2007-10-23T20:57:40  <dreimark> bb 2mins
2007-10-23T20:57:57  <dreimark> ThomasWaldmann: have to log in
2007-10-23T21:01:12  <dreimark> bb
2007-10-23T21:49:13  <dreimark> comments action link used on RC has strange effects
2007-10-23T21:50:00  <ThomasWaldmann> hehe
2007-10-23T21:50:24  <ThomasWaldmann> it hides and shows class comment elements
2007-10-23T21:51:19  <dreimark> ir resizes the list it gots compressed
2007-10-23T21:56:59  <ThomasWaldmann> the main problem is the day bar not returning to its original width
2007-10-23T23:19:24  <dreimark> gn
2007-10-23T23:20:00  <xorAxAx> gn dreimark