MoinMoin: MoinMoinChat/Logs/moin-dev/2008-06-14

2008-06-14T00:19:34  <dreimark> re
2008-06-14T00:20:24  <dreimark> http://moinmo.in/MoinMoinBugs/ErrorOnAttachZipFileToAPageWhenSubdirectoriesAreInTheZipFile?action=AttachFile&do=unzip&target=TestWiki.zip
2008-06-14T00:20:48  <dreimark> ThomasWaldmann: is that correct to get a RuntimeError ?
2008-06-14T00:47:23  <dreimark> good night
2008-06-14T01:01:36  * ThomasWaldmann looks
2008-06-14T01:08:03  <ThomasWaldmann> dreimark: yeah, another place that needs try/except
2008-06-14T01:11:55  <ThomasWaldmann> dreimark: where was that url shown by moin?
2008-06-14T01:49:43  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3702:b6dcdf55795e 1.7/MoinMoin/action/AttachFile.py: AttachFile action: catch RuntimeError when someone tries to unzip a defective archive
2008-06-14T08:27:23  <dreimark> ThomasWaldmann: I have had it in my firefox url list from some of the previous trials
2008-06-14T08:27:55  <dreimark> it cames into foreground yesterday because of recovering a firefox session
2008-06-14T11:17:58  <ThomasWaldmann> moin
2008-06-14T11:18:21  <ThomasWaldmann> dreimark: but the moin UI does not generate that url any more, right?
2008-06-14T11:25:04  <PawelPacana> johill: dennda: create_item should not create first Item revision, we let this for item.commit() right?
2008-06-14T11:26:44  <dreimark> ThomasWaldmann: yes, you don't have the dialog but if you know the do parameters you can still use them
2008-06-14T11:26:56  <CIA-51> Pawel Pacana <pawel.pacana@gmail.com> default * 4031:14a7a398787e 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py:
2008-06-14T11:26:56  <CIA-51> Implemented _list_revisions method for listing available Item revisions. Updated
2008-06-14T11:26:56  <CIA-51> docstring.
2008-06-14T11:26:57  <CIA-51> Pawel Pacana <pawel.pacana@gmail.com> default * 4032:db424e551966 1.8-mercurialbackend-ppacana/MoinMoin/storage/ (abstract.py backends/hg.py): Added Revision class stub for compatibility.
2008-06-14T11:27:15  <dreimark> ThomasWaldmann: I think we should stop that
2008-06-14T11:27:39  <ThomasWaldmann> dreimark: see my last push
2008-06-14T11:28:39  <dreimark> seen already, and tried it with tha package file there
2008-06-14T11:29:16  <dreimark> this file is correct and I can unzip it and I don't get the unzip link shown regular because it is a  package
2008-06-14T11:29:55  <dreimark> so I think unzip of package files should be blocked too
2008-06-14T11:31:36  <dreimark> I mean we should not only block the ui we should block the action too
2008-06-14T11:32:50  <ThomasWaldmann> why should we block that?
2008-06-14T11:35:20  <dreimark> because to be consistent
2008-06-14T11:35:29  <dreimark> e.g. _do_move
2008-06-14T11:35:39  <dreimark>  if not request.user.may.delete(pagename):
2008-06-14T11:35:40  <dreimark>         return _('You are not allowed to move attachments from this page.')
2008-06-14T11:38:37  <dreimark> if we exclude unzip from the ui because of a page then noone should be able to unzip it by changing the url
2008-06-14T11:38:58  <johill> PawelPacana: yeah
2008-06-14T11:39:15  <dreimark> s/because of a page/because of a package/
2008-06-14T11:40:09  <ThomasWaldmann> it is a difference to not show something because it would crash the whole page or not allowing something because we think it is no good doing it
2008-06-14T11:40:55  <dreimark> well we have removed the unzip from the ui for the systempagessetup
2008-06-14T11:41:19  <dreimark> because one with no superuser rights was able to unzip the lang pack
2008-06-14T11:42:18  <ThomasWaldmann> that's silly
2008-06-14T11:42:35  <dreimark> it is
2008-06-14T11:42:57  <ThomasWaldmann> if you allow users to unzip files, they can zip together arbitrary stuff and unzip it
2008-06-14T11:43:25  <ThomasWaldmann> so why should we check for and disallow unzipping packages?
2008-06-14T11:43:38  <dreimark> because we do it for the ui
2008-06-14T11:43:55  <dreimark> my point is it makes sense to forbide it in the ui if it is a package
2008-06-14T11:45:41  <ThomasWaldmann> not sure about that either
2008-06-14T11:46:11  <dreimark> you don't want in attachments files enumbered by 1 to a few hundred
2008-06-14T11:46:24  <dreimark> without any usage
2008-06-14T11:48:36  <dreimark> and I don't want to be the one who has to purge them from the attach file list using the ui
2008-06-14T11:50:00  <ThomasWaldmann> that might be true, but that is a general problem and not limited to packages
2008-06-14T11:50:36  <ThomasWaldmann> a user can upload a zip with foo-1 .. foo-1000 zero byte content
2008-06-14T11:50:59  <ThomasWaldmann> if this is within limits, he is allowedunzipping it
2008-06-14T11:51:39  <dreimark> that's true
2008-06-14T11:52:09  <ThomasWaldmann> (or some other user is allowed too, not knowing what exactly he unzips there)
2008-06-14T11:52:09  <dreimark> unzipping a package is done by accident easy if the link is in the ui
2008-06-14T11:52:56  <gizmach> re
2008-06-14T11:54:12  <dreimark> for those problems later xapian index build and may be spambayes weihting could help to identify those files
2008-06-14T11:54:16  <dreimark> ThomasWaldmann:
2008-06-14T11:54:36  <dreimark> s/weihting/weighting/
2008-06-14T11:56:26  <ThomasWaldmann> or just asking 'do you really want to unzip 1000 files'
2008-06-14T11:56:47  <ThomasWaldmann> or not allowing someone to unzip 1000 files by using much lower limits
2008-06-14T11:57:48  <ThomasWaldmann> this whole stuff will have to be reconsidered anyway after we have new storage backend
2008-06-14T11:58:29  <ThomasWaldmann> hg rm AttachFile.py will solve a lot of problems :)
2008-06-14T11:58:52  <dreimark> yeah, we should not change there to much now otherwise johill has to merge again ;)
2008-06-14T12:02:23  <dreimark> btw. using the tab feature makes it simple to del attachments
2008-06-14T12:02:56  <dreimark> (right mouse click, open in new tab)
2008-06-14T12:04:08  <ThomasWaldmann> no, you just use the middle button of ff :)
2008-06-14T12:06:48  <dreimark> middle wheel button
2008-06-14T12:08:08  <dreimark> bbl
2008-06-14T12:55:56  <dennda> johill: yes, just got reincarnated
2008-06-14T12:56:00  <dennda> a bit difficult
2008-06-14T12:56:23  <dennda> taking a shower now and some work afterwards :)
2008-06-14T13:17:53  <johill> :)
2008-06-14T13:20:27  <dreimark> johill:  I do have a question on superuser change
2008-06-14T13:21:04  <dreimark> if I su do a user is it right that I get shown the selectuser dialog from that user
2008-06-14T13:21:42  <dreimark> without that this user has su previleges
2008-06-14T13:21:46  <johill> yeah, the select user dialog takes the superuser setting
2008-06-14T13:21:54  <johill> so you can switch to another one without logging out
2008-06-14T13:22:03  <dreimark> ok
2008-06-14T13:23:43  <dreimark> and I guess it is right that I do see always my page trail and not the one of the user I have su'ed to
2008-06-14T13:24:01  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3703:95ed56263121 1.7/MoinMoin/i18n/ (7 files): updated i18n
2008-06-14T13:25:07  * dreimark checks currently trail for fcgi
2008-06-14T13:26:35  <dreimark> a feature for SystemInfo could be to show the deployment method
2008-06-14T13:29:04  <ThomasWaldmann> dreimark: soon it is all wsgi anyway
2008-06-14T13:29:55  <dreimark> :)
2008-06-14T13:29:57  <ThomasWaldmann> anything left we need to fix before rc3?
2008-06-14T13:30:20  <dreimark> mitsuhiko: do you have contact to chris arndt? Is he well?
2008-06-14T13:31:03  <mitsuhiko> dreimark: no contact. sorry
2008-06-14T13:33:20  <dreimark> ThomasWaldmann: moment
2008-06-14T13:36:15  <dreimark> ThomasWaldmann: anon user does not get a session with default config, that's true or ?
2008-06-14T13:40:45  <johill> dreimark: yeah true
2008-06-14T13:40:58  <dreimark> ok
2008-06-14T13:41:24  <dreimark> the next question is about this autologin we discussed for ldap a few days ago
2008-06-14T13:41:45  <dreimark> should it log me in now or not after creating a user?
2008-06-14T13:41:48  <dreimark> johill:
2008-06-14T13:42:55  <johill> dunno, doesn't really matter
2008-06-14T13:44:39  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3704:8ffe79f29100 1.7/ (docs/INSTALL.html wiki/data/intermap.txt docs/UPDATE.html): make install-docs interwiki, add UPDATE.html
2008-06-14T13:44:55  <dreimark> it does not, I do prefer it this way because sometimes a secretary does create accounts
2008-06-14T13:45:40  <gizmach> is there any way currently to distinguish usergroups and pagegroups? ( i know that member of a group is for example GroupMember and I can't distingush it) right?
2008-06-14T13:47:06  <dreimark> currently not they are all page_group_regex.
2008-06-14T13:47:08  <dreimark> gizmach:
2008-06-14T13:47:18  <gizmach> :)
2008-06-14T13:47:19  <gizmach> ok
2008-06-14T13:47:56  <ThomasWaldmann> gizmach: why does it matter?
2008-06-14T13:48:05  <gizmach> I'm just thinking which is the best/easiest way to split them
2008-06-14T13:48:34  <dreimark> e.g. SystemPagesInEnglishGroup won't never be used for acls
2008-06-14T13:48:51  <dreimark> but it is a common group page on the wiki
2008-06-14T13:50:48  <dreimark> ThomasWaldmann: later on you would like to access those by page_group_regex and you do want user_groups from ldap
2008-06-14T13:51:53  <kikkay> Hello
2008-06-14T13:52:33  <ThomasWaldmann> "differentiating between user groups and other groups does matter, because the following won't work if you don't differentiate: ..."
2008-06-14T13:52:42  <ThomasWaldmann> please complete that sentence :)
2008-06-14T13:53:15  <ThomasWaldmann> hi kikkay
2008-06-14T13:53:21  <johill> you need to take a communication class too ;)
2008-06-14T13:56:26  <ThomasWaldmann> gizmach: ?
2008-06-14T13:56:39  <dreimark> ThomasWaldmann: do you assume to have all group pages you currently have in moin will be on the ldap server ?
2008-06-14T13:56:48  <ThomasWaldmann> no
2008-06-14T13:57:04  <johill> fwiw, I think I agree with Thomas, you only need to look up a group on the ldap server when it's used in an acl
2008-06-14T13:57:36  <johill> but he should have asked why you need to distinguish instead. matter of communication :)
2008-06-14T13:58:01  <dreimark> no you agree with us, that is the way how it is destinguished
2008-06-14T13:58:20  <dreimark> if a group used in acls it is the common spoken "usergroup"
2008-06-14T13:58:47  <dreimark> johill:
2008-06-14T13:58:49  <johill> yeah so why do you need to know up-front whether it's a user or page group?
2008-06-14T13:59:54  <ThomasWaldmann> the only thing you want to know for acls is "u.name in aclgroup"
2008-06-14T14:00:46  <ThomasWaldmann> and for quite some other application you want to check "something in group"
2008-06-14T14:01:04  <ThomasWaldmann> so what you have to implement is some sane group membership check
2008-06-14T14:01:22  <ThomasWaldmann> and it does not matter at all what kind if group it is
2008-06-14T14:01:51  <ThomasWaldmann> groups are just collections of unicode objects
2008-06-14T14:03:37  <ThomasWaldmann> gizmach: still there? you are so silent... :)
2008-06-14T14:04:23  <gizmach> ThomasWaldmann:  I'm here, just reading, thinking and looking at wikidicts.py
2008-06-14T14:04:58  <dreimark> johill: it is not up front meant it is just that you need to destinguish at some point that this group is used for acls
2008-06-14T14:05:20  <gizmach> also I'm not sure that id doesn't matter what kind of group it i
2008-06-14T14:05:35  <gizmach> because on ldap you will have only user groups
2008-06-14T14:05:48  <ThomasWaldmann> dreimark: the acl itself determines what it uses, not the group definition
2008-06-14T14:06:01  <gizmach> ThomasWaldmann: ok that was the answer for me too
2008-06-14T14:06:02  <gizmach> :=
2008-06-14T14:06:05  <gizmach> :)
2008-06-14T14:07:31  <ThomasWaldmann> we won't hold somebody back putting NonuserGroup into an acl. it makes no sense, of course but we don't need to deal with whether it makes sense.
2008-06-14T14:08:08  <ThomasWaldmann> (that's the reason why you don't give admin rights to idiots :)
2008-06-14T14:08:29  <gizmach> :)
2008-06-14T14:09:31  <ThomasWaldmann> (there is also nothing holding somebody back from adding some bullshit to a group page that contained only valid users before he edited it)
2008-06-14T14:10:45  <ThomasWaldmann> (he could add an invalid user, a nonexisting user, a page name, unsupported wiki markup, ... - that's why it is good to have not everybody editing your group pages, but only people you trust to do it in a sane way)
2008-06-14T14:11:01  <dreimark> well then I got an idea for a feature request. I like to ignore group definitions on MM for acls if an ldap server is in place.
2008-06-14T14:11:14  <ThomasWaldmann> no
2008-06-14T14:11:18  <gizmach> :)
2008-06-14T14:11:34  <ThomasWaldmann> as we do it for auth, there should be some lookup order
2008-06-14T14:12:06  <ThomasWaldmann> so you can determine it e.g. looks first in ldap, then in the wiki, then in ...
2008-06-14T14:12:09  <gizmach> mean like for auth check if ldap is on then use ldap groups?
2008-06-14T14:12:11  <gizmach> :)
2008-06-14T14:12:35  <dreimark> and what does happen if you only have ldap enabled ? or do we assume one has ldap + moin_groups
2008-06-14T14:12:49  <dreimark> + moin_groups always ?
2008-06-14T14:12:55  <ThomasWaldmann> we use what the admin configured
2008-06-14T14:13:45  <ThomasWaldmann> that could be ldap groups only, or ldap groups + wiki groups, or ldap + config + wiki, or ...
2008-06-14T14:14:07  <ThomasWaldmann> that's why i am telling that it should start with wikidicts refactoring
2008-06-14T14:14:18  <mitsuhiko> dreimark: do you know him?
2008-06-14T14:14:18  <dreimark> I am telling this too
2008-06-14T14:14:26  <ThomasWaldmann> splitting it in highlevel code and backend code
2008-06-14T14:14:43  <ThomasWaldmann> then you'll need some way to configure backends
2008-06-14T14:14:57  <ThomasWaldmann> and some way to define what is used, in what order etc.
2008-06-14T14:15:13  <dreimark> mitsuhiko: yes, he is the founder of the pycologne user group, and I do meet him often there
2008-06-14T14:15:48  <mitsuhiko> ah
2008-06-14T14:16:03  <dreimark> mitsuhiko: see query
2008-06-14T14:17:00  <ThomasWaldmann> after doing this refactoring, you will have to think about how your ldap backend works
2008-06-14T14:17:13  <gizmach> ThomasWaldmann: ok
2008-06-14T14:17:29  <ThomasWaldmann> it could either do some cheap realtime checks (member in group checks).
2008-06-14T14:18:14  <ThomasWaldmann> or, ldap login could query all groups this user is member in and cache it to session storage at login time
2008-06-14T14:18:38  <ThomasWaldmann> the backend would then just check the session for the member in group check
2008-06-14T14:20:00  <ThomasWaldmann> i guess the last thing is cheaper because it only does one bigger ldap query for the session lifetime
2008-06-14T14:20:34  <ThomasWaldmann> while the first method might do 1 smaller ldap query per http request of any user in the worst case
2008-06-14T14:21:30  <ThomasWaldmann> hmm, no, the worst case is worse :) if a user requests TitleIndex, it will have to do lots of ACL checks...
2008-06-14T14:22:08  <johill> querying up-front is wrong though
2008-06-14T14:22:14  <johill> sessions can live very long
2008-06-14T14:22:20  <johill> and groups can change
2008-06-14T14:22:34  <ThomasWaldmann> this method is also used by windows
2008-06-14T14:22:39  <johill> "premature optimisation is the root of all evil" ;)
2008-06-14T14:22:49  <ThomasWaldmann> and the admin can have shorter sessions by configuration
2008-06-14T14:23:01  <johill> nope, he can't
2008-06-14T14:23:09  <ThomasWaldmann> ?
2008-06-14T14:23:21  <johill> if the session lifetime is > 10 minutes and the user requests a pageg every 9 minues, his session will last forever
2008-06-14T14:23:39  <ThomasWaldmann> ok, then we need a max lifetime
2008-06-14T14:23:51  <johill> get a grip
2008-06-14T14:24:00  <johill> you don't even know yet whether that ldap search will be a problem
2008-06-14T14:24:17  <johill> and if it is you can cache it differently
2008-06-14T14:24:18  <ThomasWaldmann> i know moin does very much acl checking
2008-06-14T14:24:50  <ThomasWaldmann> and a wiki can have many users doing many requests (title index, search, page lists)
2008-06-14T14:25:07  <johill> I'm not saying you don't have to cache it
2008-06-14T14:25:17  <johill> all I'm saying is that doing it in the session makes no sense
2008-06-14T14:25:24  <johill> and complicating the session code for that makes even less sense
2008-06-14T14:25:33  <ThomasWaldmann> if you cache it, you have exactly the same problem: you have to define some lifetime
2008-06-14T14:25:35  <johill> besides
2008-06-14T14:25:46  <johill> it may not even be possible to cheaply determine "what groups is this user in"
2008-06-14T14:26:28  <johill> sure, but the problem is much more defined, "cache lifetime of ldap group cache"
2008-06-14T14:26:33  <ThomasWaldmann> pyldap author told me windows does that on login
2008-06-14T14:26:35  <johill> rather than "session lifetime because it's used for ldap cache"
2008-06-14T14:27:30  <johill> and you don't think that maybe non-windows ldap installations exist, or maybe somebody will want to extend this further to non-ldap/non-moin group storage?
2008-06-14T14:27:34  <johill> that's pretty short-sighted if you ask me
2008-06-14T14:28:16  <ThomasWaldmann> i am just talking about ldap backend here, not telling that other backends should do the same
2008-06-14T14:28:17  <dreimark> however I think it should be configurabel to switch the cache for ldap off
2008-06-14T14:28:32  <dreimark> and to start with no cache,
2008-06-14T14:28:43  <johill> dreimark: sane approach, make it work first, then think about optimisations
2008-06-14T14:30:22  <gizmach> huh
2008-06-14T14:31:12  <dennda> Ok trying to get back into this
2008-06-14T14:31:58  <dennda> johill: Just wondering -  Is the metadata class actually needed?
2008-06-14T14:32:10  <johill> which one?
2008-06-14T14:32:17  <dennda> (I shall go get a little notepad. maybe we already discussed that)
2008-06-14T14:32:35  <dennda> johill: class Metadata(object) -- right at the bottom of storage.txt
2008-06-14T14:33:20  <johill> oh
2008-06-14T14:33:27  <johill> yeah, it's not
2008-06-14T14:33:32  <dennda> ok
2008-06-14T14:37:27  <dreimark> ThomasWaldmann: the windows ldap admin I spoke to does not use ldap directly for giving group rights
2008-06-14T14:37:58  <dreimark> they do use it together with ntfs filesystem rights
2008-06-14T14:38:17  <dreimark> he explained me that doing it this way is the common way MS told to do it
2008-06-14T14:39:24  <ThomasWaldmann> i don't deal with such stuff
2008-06-14T14:39:49  <ThomasWaldmann> that why I was recommending melita and you to discuss that on openldap mailing list.
2008-06-14T14:40:43  <ThomasWaldmann> so that something practically useful/usable is going to be implemented
2008-06-14T14:53:39  <ThomasWaldmann> btw, midterm evaluation is in 1 month from now
2008-06-14T14:55:19  <gizmach> ThomasWaldmann: :)
2008-06-14T14:57:53  <dennda> johill: Do we have a concrete plan already on how to handle actual (binary) Revision data?
2008-06-14T14:58:08  <dennda> In storage.txt you suggest using file-like access
2008-06-14T14:58:18  <dennda> but that suggestion is marked with a questionmark :)
2008-06-14T14:58:36  <dennda> I feel that's a reasonable way to do it but want your opinion
2008-06-14T15:05:15  <johill> dennda: I don't think we have a plan
2008-06-14T15:05:41  <dennda> Well, what options do we have?
2008-06-14T15:05:44  <johill> it makes sense to not load the whole thing into memory for large files though, so file-like would be a good thing at least where possible (i.e. I'm not sure sql blobs can be read like that)
2008-06-14T15:05:56  <dennda> yes
2008-06-14T15:05:59  <dennda> true
2008-06-14T15:06:13  <dennda> But that's a limitation of the specific backend
2008-06-14T15:06:17  <johill> but even then you can of course just use StringIO for file-like
2008-06-14T15:06:31  <johill> well, not exactly, because it must not be writable, but anyway
2008-06-14T15:06:47  <dennda> Only if SQL supports reading only a chunk of bytes
2008-06-14T15:06:51  <dennda> I don't know if that is possible
2008-06-14T15:06:59  <johill> yeah I wouldn't worry about it
2008-06-14T15:07:21  <johill> file-like is the best option, let the backend handle the rest
2008-06-14T15:07:45  <dennda> I agree
2008-06-14T15:07:49  <dennda> which leads to the next question
2008-06-14T15:08:03  <dennda> How do you implement that? Must one do it all by hand or is there a nice shortcut?
2008-06-14T15:08:13  <johill> I don't know of any shortcut
2008-06-14T15:08:19  <johill> but isn't it basically just a .read() method?
2008-06-14T15:08:24  <johill> and maybe seek?
2008-06-14T15:08:33  <johill> seek shouldn't even be necessary
2008-06-14T15:08:54  <dennda> how do you fill in data? you pass it when you create the revision I assume
2008-06-14T15:08:59  <dennda> thus, no need for write()
2008-06-14T15:09:18  <johill> oh yeah for newly created revisions you want write
2008-06-14T15:09:28  <johill> so you can stream the data right from the browser into the backend
2008-06-14T15:09:33  <johill> for uploads
2008-06-14T15:10:01  <johill> the other  thing files have is 'for line in f:', not sure if we need that
2008-06-14T15:10:10  <dennda> don't think so
2008-06-14T15:10:13  <johill> anyway, back in a bit, tea time
2008-06-14T15:10:18  <dennda> you only need that for plaintext files, don't you?
2008-06-14T15:10:18  <dennda> ok
2008-06-14T15:15:24  <dreimark> bbl
2008-06-14T15:26:48  <dennda> johill: forget what I just said, please
2008-06-14T15:27:19  <dennda> johill: Item and Revision use metadata the same way. I feel we should create a seperate metadata(DictMixin) class and let Item and Revision inherit from it
2008-06-14T15:28:09  <ThomasWaldmann> http://moinmo.in/4ct10n/diff/GoogleSoc2008?action=diff&date=1213449148000000 < all mentors/students please read and add it to your schedule
2008-06-14T15:29:20  <johill> back
2008-06-14T15:29:31  <johill> dennda: sure
2008-06-14T15:30:03  <johill> ThomasWaldmann: it's very unlikely that I can be online Friday 12-1
2008-06-14T15:30:15  <ThomasWaldmann> (I am open to discuss the weekday/time, but we have to be aware that there likely is no time working for everybody every week)
2008-06-14T15:30:50  <ThomasWaldmann> and please convert to localtime yourself
2008-06-14T15:30:54  <dennda> my fridays are free most of the time
2008-06-14T15:31:21  <johill> well yeah, that's 2-3pm where I'm usually sitting on a train
2008-06-14T15:31:38  <johill> or in the process of getting there or similar
2008-06-14T15:33:04  <johill> besides
2008-06-14T15:33:28  <johill> is it really useful if everybody does this at the same time?
2008-06-14T15:33:41  <dennda> oh
2008-06-14T15:33:52  <dennda> Now I got Revision(Metadata) and Item(Metadata)
2008-06-14T15:33:55  <johill> I personally don't really care about what waldi is doing
2008-06-14T15:34:00  <dennda> but that's not a "is-a" relationship
2008-06-14T15:34:00  <ThomasWaldmann> i guess it is easier for having comments from all mentors
2008-06-14T15:34:01  <johill> for instance
2008-06-14T15:34:26  <ThomasWaldmann> (you can discuss with your students at any time anyway)
2008-06-14T15:34:38  <ThomasWaldmann> this is more a common thing
2008-06-14T15:34:46  <johill> right, but why should I be interested?
2008-06-14T15:35:12  <johill> dennda: yeah, that's true, also, the access rules are quite different, so maybe it doesn't make that much sense
2008-06-14T15:35:38  <ThomasWaldmann> obviously you should be interested e.g. in the mercurial backend :)
2008-06-14T15:35:53  <johill> well yeah but then I'm interested in a third of the discussion
2008-06-14T15:36:22  <ThomasWaldmann> we won't force you to read or comment the rest :)
2008-06-14T15:36:38  <johill> then why dictate a common time?
2008-06-14T15:36:41  <dennda> johill: access rules?
2008-06-14T15:36:51  <johill> dennda: yeah like when things are writable
2008-06-14T15:37:30  <dennda> johill: the only methods I pulled out are getitem, setitem lock and unlock
2008-06-14T15:37:53  <johill> but revisions don't lock/unlocak
2008-06-14T15:37:56  <dennda> the other option is to have code duplication to some extent
2008-06-14T15:37:57  <johill> -a
2008-06-14T15:38:00  <ThomasWaldmann> having a definite time makes it clearer for the students. they know that at that time their progress will be discussed, they will declare their blockers, they will get input about how to unblock.
2008-06-14T15:38:20  <ThomasWaldmann> and about how to proceed.
2008-06-14T15:48:22  <johill> dennda: yes, but there won't be all that much code duplicated, and it probably makes more sense that way because of the different writable rules
2008-06-14T15:49:19  <johill> you're spending far too much time on exclusive API design imho, you really should start writing that memory backend too
2008-06-14T15:51:53  <ThomasWaldmann> memory backend?
2008-06-14T15:52:13  <johill> just very basic in-memory storage
2008-06-14T15:52:21  <johill> for API tests
2008-06-14T15:52:55  <dreimark> johill: students can learn from each other too
2008-06-14T15:52:59  * ThomasWaldmann .oO(fastest backend ever, but you need a very good UPS :)
2008-06-14T15:53:20  <dreimark> no not really did you read the princeton article?
2008-06-14T15:53:51  <ThomasWaldmann> me?
2008-06-14T15:56:42  <dreimark> http://citp.princeton.edu/memory/
2008-06-14T15:57:23  <dreimark> ThomasWaldmann: Cold Boot Attacks on Encryption Keys
2008-06-14T15:57:39  <johill> was that really news to anyone?
2008-06-14T15:57:55  <johill> I've often enough cold-booted my machine and seen things in the graphics memory that I had open when it crashed
2008-06-14T15:59:01  <johill> it struck me as odd that people were so irritated, the effect itself didn't seem noteworthy to me (maybe the fact that they successfully used it, but still)
2008-06-14T15:59:20  <dreimark> not really new but how they tell it
2008-06-14T15:59:59  <ThomasWaldmann> dreimark: still unlikely that would be useful for moin memory backend :)
2008-06-14T16:00:06  <dreimark> hehe
2008-06-14T16:20:58  <dennda> ok
2008-06-14T16:24:01  <johill> dennda: oh and were you going to put it into __init__.py?
2008-06-14T16:24:29  <dennda> the whole abstract thing?
2008-06-14T16:25:01  <johill> yeah
2008-06-14T16:25:07  <dennda> I can, yes
2008-06-14T16:32:31  <johill> I just think it makes for nicer imports
2008-06-14T16:33:37  <zenhase> re
2008-06-14T17:41:54  <dreimark> bbl
2008-06-14T17:58:31  <ThomasWaldmann> johill: what was the reason again to keep order in the groups?
2008-06-14T18:00:26  <johill> ThomasWaldmann: for page groups, I use it to automatically build sitemaps and stuff like that in the theme
2008-06-14T18:00:43  <johill> so you can edit short-cut links in the wiki
2008-06-14T18:00:45  <johill> like navibar
2008-06-14T18:00:49  <johill> but in the wiki instead of config
2008-06-14T18:00:54  <ThomasWaldmann> ah, ok
2008-06-14T18:00:58  <ThomasWaldmann> gizmach: ^^
2008-06-14T18:01:23  <johill> that way I don't have to be bothered by the users every time they want to change the navigation a bit :)
2008-06-14T18:01:59  <ThomasWaldmann> looks like in fact we have list, set and dict usage
2008-06-14T18:02:08  <johill> yeah
2008-06-14T18:02:16  <johill> initially I actually added a WikiList class
2008-06-14T18:02:40  <johill> but then somebody complained and I just made the normal stuff keep order
2008-06-14T18:05:17  <ThomasWaldmann> hmm, the question is if it is worth having 3 separate classes for that or use some "ordered dict" for all of them
2008-06-14T18:05:55  <johill> yeah but ldap groups won't be ordered. of course, there it also doesn't matter
2008-06-14T18:07:36  <ThomasWaldmann> gizmach: think about it. currently it is all "ordered dict", but you can change that.
2008-06-14T18:09:36  <ThomasWaldmann> (the current Group.addmember code is obviously not perfect for adding a already existing member)
2008-06-14T18:20:39  <gizmach> re
2008-06-14T18:21:48  <gizmach> ok,..
2008-06-14T18:24:14  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3705:01ef230fb671 1.7/ (MoinMoin/version.py docs/CHANGES moin.spec): 1.7.0rc3 release (bump version number, update CHANGES, ...)
2008-06-14T18:24:17  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3706:cfc8d184f884 1.7/.hgtags: tagged release 1.7.0rc3
2008-06-14T18:24:56  <johill> release on the 17th? ;)
2008-06-14T18:25:53  <ThomasWaldmann> not sure :)
2008-06-14T18:32:53  <ThomasWaldmann> hah, bug:
2008-06-14T18:33:07  <ThomasWaldmann> Benachrichtigungen versendet an: srivag,DjahanMoharregh,BradeyHonsinger,simple,...(very long unbreakable list)
2008-06-14T18:33:17  <johill> heh
2008-06-14T18:33:23  <johill> ','.join instead of ', '.join?
2008-06-14T18:33:29  <ThomasWaldmann> looks like
2008-06-14T18:34:02  <johill> I haven't seen teh double-notification again since I added the backtrace info to it
2008-06-14T18:34:14  <ThomasWaldmann> heisenbug :)
2008-06-14T18:36:35  <johill> yeah
2008-06-14T18:43:34  <CIA-51> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 3707:78f1b1c0affb 1.7/MoinMoin/PageEditor.py: PageEditor: fixed msg displaying the notified people - use a breakable list
2008-06-14T19:15:33  <mmihaljevic> damn
2008-06-14T19:23:04  <TheSheep> mmihaljevic: exams?
2008-06-14T19:23:39  <mmihaljevic> TheSheep: all except one passed
2008-06-14T19:23:40  <mmihaljevic> gtg
2008-06-14T19:41:14  <kikkay> moin
2008-06-14T19:49:00  <ThomasWaldmann> hi kikkay
2008-06-14T19:53:59  <kikkay> Hey ThomasWaldmann
2008-06-14T20:04:48  <kikkay> huh, where is zenhase?
2008-06-14T20:07:13  <ThomasWaldmann> zenhase: where are you? :)
2008-06-14T20:08:10  <kikkay> huch.?
2008-06-14T20:08:32  <kikka> :( stupid client.
2008-06-14T22:33:42  <CIA-51> Marian Neagul <marian@segfault.ro> default * 4001:27666ec0b111 1.7-classify-mneagul/docs/CHANGES: Merge with main
2008-06-14T22:33:43  <CIA-51> Marian Neagul <marian@segfault.ro> default * 4002:2123615e0bc0 1.7-classify-mneagul/MoinMoin/classifier/spam_bayes_rpc.py: Fix invalid remote action
2008-06-14T22:33:44  <CIA-51> Marian Neagul <marian@segfault.ro> default * 4003:cba4750797b9 1.7-classify-mneagul/MoinMoin/action/__init__.py: Change the way we display the spam notification
2008-06-14T22:33:45  <CIA-51> Marian Neagul <marian@segfault.ro> default * 4004:865576fccbe8 1.7-classify-mneagul/MoinMoin/action/togglespamstatus.py: Change the dialog display method
2008-06-14T22:33:47  <CIA-51> Marian Neagul <marian@segfault.ro> default * 4005:f38ff866b22f 1.7-classify-mneagul/MoinMoin/classifier/__init__.py: Change Group code, needs fixing.