MoinMoin: MoinMoinChat/Logs/moin-dev/2009-05-28

2009-05-28T00:29:37  *** grzywacz has quit IRC
2009-05-28T00:40:14  *** amartani has joined #moin-dev
2009-05-28T00:57:21  <amartani> moin
2009-05-28T01:25:24  <ThomasWaldmann> re
2009-05-28T01:25:28  <ThomasWaldmann> hi amartani
2009-05-28T01:26:20  <ThomasWaldmann> are you making progress?
2009-05-28T01:45:17  <amartani> ThomasWaldmann: some.. I have some questions about 2.0...
2009-05-28T01:45:51  <ThomasWaldmann> yes?
2009-05-28T01:46:59  <amartani> Is it planned to support more than one editor for the same mimetype (as the editor is now in the Item classes)?
2009-05-28T01:48:07  <ThomasWaldmann> yes, we have 2 editors in 1.9 and we plan to have also multiple in 2,0
2009-05-28T01:50:51  <amartani> hum, I'm doing my changes in the main text editor, for now.. should I work on support multiple editors first?
2009-05-28T01:51:11  <ThomasWaldmann> can be added later
2009-05-28T01:57:25  <amartani> ok... how should I link the static files from the editor template? (I'm hardcoding paths now..)
2009-05-28T01:59:18  <ThomasWaldmann> there is request.cfg.url_prefix_local (or similar), use that as prefix
2009-05-28T02:00:48  <ThomasWaldmann> if it is unclear to you how to use it, then see PageGraphicalEditor.py in moin/1.9 repo
2009-05-28T02:03:15  <amartani> ok, thanks..
2009-05-28T02:31:17  <amartani> how can I catch every request to, say, /mobwrite?
2009-05-28T03:11:22  <CIA-19> Alexandre Martani <amartani AT gmail DOT com> default * 5791:bd4996dd243b 2.0-storage-editor-amartani/MoinMoin/ (items/__init__.py templates/modify_text.html): Real-time editor: Use url_prefix to include static files instead of hardcoded
2009-05-28T03:21:29  *** rkerr has quit IRC
2009-05-28T03:21:46  *** rkerr has joined #moin-dev
2009-05-28T03:27:47  *** ser has quit IRC
2009-05-28T03:27:56  *** ser has joined #moin-dev
2009-05-28T04:06:30  *** amartani has quit IRC
2009-05-28T04:10:13  *** amartani has joined #moin-dev
2009-05-28T04:17:53  <CIA-19> Alexandre Martani <amartani AT gmail DOT com> default * 5792:57bec1cb47d5 2.0-storage-editor-amartani/contrib/GoogleMobwrite/ (diff_match_patch.py mobwrite_core.py mobwrite_daemon.py): Real-time editor: Add google mobwrite daemon
2009-05-28T04:24:36  *** amartani has quit IRC
2009-05-28T06:55:20  <dimazest> moin
2009-05-28T08:18:14  <ThomasWaldmann> moin
2009-05-28T08:19:27  *** PawelPacana has quit IRC
2009-05-28T08:20:18  *** PawelPacana has joined #moin-dev
2009-05-28T13:33:54  <ser> TheSheep: ping
2009-05-28T13:46:32  <dreimark> moin
2009-05-28T14:20:29  <TheSheep> ser: ack
2009-05-28T14:36:29  <dreimark> anyone seen RC toggling a page New/Deleted ?
2009-05-28T15:53:01  <ThomasWaldmann> no
2009-05-28T16:06:07  <dreimark> what is wrong with
2009-05-28T16:06:09  <dreimark> actions_excluded = DefaultConfig.actions_excluded
2009-05-28T16:06:15  <dreimark> actions_excluded.remove('xmlrpc')
2009-05-28T16:06:30  <dreimark> or why is that nor
2009-05-28T16:06:39  <dreimark> t shown in WikiConfig
2009-05-28T16:08:29  <dreimark> macro#
2009-05-28T16:28:05  <ThomasWaldmann> that should work, but it modifies the list in DefaultConfig
2009-05-28T18:36:04  *** grzywacz has joined #moin-dev
2009-05-28T18:42:12  <rkerr> ThomasWaldmann: some details on sreg and teams extensions http://moinmo.in/RowanKerr/OpenIDUpdates
2009-05-28T18:56:37  <dennda> how do you define a method that is called *before all other testcases* and which can access the request object created for the tests?
2009-05-28T18:58:12  <TheSheep> setUp?
2009-05-28T19:04:43  <ThomasWaldmann> rkerr: yeah, I have read it
2009-05-28T19:06:11  <rkerr> ThomasWaldmann: any questions? :)
2009-05-28T19:07:05  <ThomasWaldmann> rkerr: you have some broken links there
2009-05-28T19:09:51  <ThomasWaldmann> rkerr: that teams stuff somehow sounds like a group backend
2009-05-28T19:10:30  <rkerr> fixed openid.net and link to Johannes' page
2009-05-28T19:13:24  <rkerr> well, it creates/updates group pages based on the openid response
2009-05-28T19:15:28  <ThomasWaldmann> yes, that's like a special group backend
2009-05-28T19:16:04  <ThomasWaldmann> dimazest's SOC project is to refactor group stuff, to make it more general (and not limited to group PAGES)
2009-05-28T19:17:40  <rkerr> cool
2009-05-28T19:17:41  <ThomasWaldmann> rkerr: btw, your page talks about "ACL pages", but I guess your mean group pages
2009-05-28T19:18:13  <rkerr> yeah, group pages..
2009-05-28T19:18:42  <ThomasWaldmann> if he is successful, we can maybe handle teams similar to ldap
2009-05-28T19:20:10  <ThomasWaldmann> for ldap, there are 2 ways: either dynamically query the ldap server as we encounter group membership conditionals, or querying all group memberships at login time and store them in some cache (AFAIK this is the way windows does it in AD setups)
2009-05-28T19:23:33  <rkerr> openid teams would have to be at login time
2009-05-28T19:26:36  <ThomasWaldmann> rkerr: btw, if you want "return None" (== "return"), it is the same as doing nothing when at the end of a function or method
2009-05-28T19:28:19  <ThomasWaldmann> rkerr: all those config items - can some of them move to the auth object initialization or are they also used elsewhere?
2009-05-28T19:31:28  <rkerr> just in the auth bits... user.auth_attribs, and the user_form_disable and user_form_remove should take care of the user prefs form
2009-05-28T19:32:27  <rkerr> except for openidrp_acl_page_postfix i suppose only because it is used for setting page_group_regex
2009-05-28T19:33:48  <ThomasWaldmann> user_form_* stuff is standard moin stuff
2009-05-28T19:33:56  <ThomasWaldmann> i rather mean the new config items
2009-05-28T19:34:47  <ThomasWaldmann> we try to keep all stuff specific to some auth class in the auth object init, without requiring some extra attribute in multiconfig or wikiconfig
2009-05-28T19:35:22  <ThomasWaldmann> for example, we move all ldap* stuff from the config to the LDAPAuth init
2009-05-28T19:35:31  <rkerr> openidrp_allowed_op, openidrp_allow_registration, and openidrp_registration_url are used outside of auth
2009-05-28T19:36:17  <dennda> diff is broken in 1.9
2009-05-28T19:36:20  <ThomasWaldmann> and the sreg stuff?
2009-05-28T19:37:29  <rkerr> sreg values are only used in the sreg extension but may vary from site to site
2009-05-28T19:39:16  <ThomasWaldmann> yes, and sreg is done within auth, right?
2009-05-28T19:40:02  <rkerr> yeah, then mapped to values in the user object
2009-05-28T19:41:14  <ThomasWaldmann> the problem with all those separate config attrs (and the reason we try to avoid that meanwhile) is that it creates a huge multiconfig.py
2009-05-28T19:41:42  <ThomasWaldmann> because everything should have some sane default, so you don't get an AttributeError if you do not define it in your wikiconfig
2009-05-28T19:42:27  <ThomasWaldmann> but since we use auth objects, all that auth related stuff can be much better defaulted when it is a __init__ param of the auth object
2009-05-28T19:42:38  <dennda> hmhm
2009-05-28T19:42:49  <rkerr> ok.. well the sreg values should be fine in auth
2009-05-28T19:43:09  <rkerr> list of teams .. not really any way to have a default for that
2009-05-28T19:43:33  <ThomasWaldmann> an empty list maybe? :)
2009-05-28T19:44:01  <ThomasWaldmann> but be careful, empty lists are no good as default argument values
2009-05-28T19:44:19  <dennda> ThomasWaldmann: got some spare minutes?
2009-05-28T19:44:27  <ThomasWaldmann> yes
2009-05-28T19:44:46  <dimazest> hi
2009-05-28T19:45:13  <dennda> ThomasWaldmann: I'd like to work a bit more test-driven; However, afaict, it's not too trivial to use the acl thing in tests as per the magic that seems to happen before tests are run
2009-05-28T19:45:46  <ThomasWaldmann> what's happening magically?
2009-05-28T19:45:55  <rkerr> well, the teams functions check for non-zero length of configured teams before adding the extension to openid requests
2009-05-28T19:46:05  <dennda> ThomasWaldmann: some "TestRequest" is made available to every testcase
2009-05-28T19:46:31  <ThomasWaldmann> yes, and what's the problem?
2009-05-28T19:46:57  <dennda> http://paste.pocoo.org/show/119591/
2009-05-28T19:47:00  <dennda> this, for instance
2009-05-28T19:47:44  * ThomasWaldmann looks
2009-05-28T19:49:04  <dennda> perhaps you can pull a solution out of your hat
2009-05-28T19:49:38  <ThomasWaldmann> looks like r.data_backend is not initialized
2009-05-28T19:49:48  <dennda> yes
2009-05-28T19:50:02  <dennda> or well, it is
2009-05-28T19:50:21  <dennda> but it's backend-attribute isn't
2009-05-28T19:51:11  <ThomasWaldmann> that is 1.9-st or 2.0-st?
2009-05-28T19:51:30  <dennda> 1.9
2009-05-28T19:51:47  <dennda> in order to be able to easily pull the tests into 2.0
2009-05-28T19:52:16  <rkerr> ThomasWaldmann: so, since the sreg and teams stuff is just a collection of functions passed to OpenIDAuth class, where would you think the config options should go? should the sreg file set new properties on OpenIDAuth class like it does for auth_attribs?
2009-05-28T19:54:11  <ThomasWaldmann> dennda: didn't I fix that stuff recently when you worked in 2.0-st?
2009-05-28T19:55:23  <dennda> ThomasWaldmann: since your fix is in 1.9 as well...
2009-05-28T19:55:25  <dennda> oh it isn't
2009-05-28T19:55:52  <dennda> thought i'd checked
2009-05-28T19:58:15  <dennda> sorry
2009-05-28T20:00:38  <ThomasWaldmann> btw, how was your exam?
2009-05-28T20:01:04  <dennda> given the short learning period I guess it was ok, but it was quite difficult
2009-05-28T20:01:12  <dennda> we'll see
2009-05-28T20:01:43  <ThomasWaldmann> topic?
2009-05-28T20:01:52  <dennda> rechnernetze
2009-05-28T20:02:32  <ThomasWaldmann> sounds easy :)
2009-05-28T20:02:54  <dennda> it's more like elektrotechnik
2009-05-28T20:05:03  * ThomasWaldmann hat fast elektrotechnik als nebenfach genommen, bis er in einer (probe)vorlesung doch ziemlich gelangweilt war.
2009-05-28T20:05:18  <dennda> geht mir auch net ab das thema
2009-05-28T20:05:22  <dennda> was juckt mich der wellenwiderstand
2009-05-28T20:06:30  <ThomasWaldmann> tja, frag mal ne Welle! :P
2009-05-28T20:09:06  <rkerr> ThomasWaldmann: would like to disable normal user registration link so users can be directed to an openid provider's registration page instead.. openidrp_allow_registration and openidrp_registration_url config vars control that but maybe they should be renamed to user_* or auth_* as they aren't really openid specific
2009-05-28T20:10:40  <ThomasWaldmann> i think our auth classes have some related stuff
2009-05-28T20:10:55  <ThomasWaldmann> iirc it is about whether or not to show login/logout link
2009-05-28T20:11:01  <ThomasWaldmann> maybe that can be just extended
2009-05-28T20:12:00  <rkerr> auth/__init__.py login_hint seems to do it
2009-05-28T20:12:29  <rkerr> so, replace login_hint with different method? or just extend it to check new options
2009-05-28T20:12:31  <dennda> I guess there aren't any acl tests I can reuse ;)
2009-05-28T20:13:09  <rkerr> i guess OpenIDAuth can re-implement it and check...
2009-05-28T20:13:11  <ThomasWaldmann> dennda: we have some ACL tests, but I think they are on another level
2009-05-28T20:13:23  <dennda> yes, if you're talking about MoinMoin.security tests
2009-05-28T20:14:16  <rkerr> err actually that won't work ... login_hint from each enabled auth class is printed
2009-05-28T20:21:36  <dennda> ok first acl violation detected ;)
2009-05-28T20:24:29  <dennda> if acl is only set to "All:read", nobody should be able to write to that page, correct?
2009-05-28T20:26:56  <ThomasWaldmann> sure
2009-05-28T20:27:14  <dennda> why does that even work in 1.9 then?
2009-05-28T20:27:20  * ThomasWaldmann is CaptainObvious :)
2009-05-28T20:27:33  <ThomasWaldmann> example?
2009-05-28T20:28:12  <ThomasWaldmann> that gdchart* stuff is ugly
2009-05-28T20:28:20  <dennda> I fired up 1.9 (non-storage), created a page, added #acl: All:read, saved, but was still able to write to it. Although the acl line got removed completely. can't I edit it because I'm not the admin of that page?
2009-05-28T20:28:42  <ThomasWaldmann> it is: #acl All:read
2009-05-28T20:29:03  <dennda> ah
2009-05-28T20:29:12  <dennda> ok, disregard what I said then
2009-05-28T20:29:46  <dennda> at least the 1.9 part. in 1.9-storage that works
2009-05-28T20:31:25  * dennda investigates that further
2009-05-28T20:46:23  <CIA-19> Rowan Kerr <rowan@stasis.org> default * 4735:6279b8badd5f 1.9-openid/MoinMoin/auth/ (3 files in 2 dirs): Removed no-value return statements from openid extensions. Prepared default auth login_prompt to have configurable registration url.
2009-05-28T20:46:48  *** devilsadvocate has quit IRC
2009-05-28T20:47:09  *** devilsadvocate has joined #moin-dev
2009-05-28T20:59:07  * ThomasWaldmann goes killing gdchart stuff from 2.0
2009-05-28T20:59:15  <dennda> ok found it
2009-05-28T21:00:35  <ThomasWaldmann> ah, I already did that
2009-05-28T21:00:49  <ThomasWaldmann> dreimark: can we remove EmbedObject from 2.0-storage?
2009-05-28T21:03:35  <ThomasWaldmann> (it doesn't work any more anyway because it uses AttachFile and that's gone)
2009-05-28T21:41:50  <CIA-19> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 5789:a5dbcaaf7194 2.0-storage/MoinMoin/macro/ (4 files in 2 dirs): (log message trimmed)
2009-05-28T21:41:50  <CIA-19> removed macros: Action and EmbedObject
2009-05-28T21:41:50  <CIA-19> Action macro was used to create action links.
2009-05-28T21:41:50  <CIA-19> This can easily be done by link syntax like now, e.g.:
2009-05-28T21:41:50  <CIA-19> [[target||&do=ACTIONNAME]]
2009-05-28T21:41:51  <CIA-19> EmbedObject was used to embed multimedia objects.
2009-05-28T21:41:55  <CIA-19> This can be done by transclusion syntax now, using {{target||PARAMS}}.
2009-05-28T22:01:29  *** rkerr1 has joined #moin-dev
2009-05-28T22:03:04  *** rkerr has quit IRC
2009-05-28T22:13:26  *** rkerr has joined #moin-dev
2009-05-28T22:29:55  *** rkerr has quit IRC
2009-05-28T22:43:34  *** rkerr1 has quit IRC
2009-05-28T22:53:54  *** amartani has joined #moin-dev
2009-05-28T22:54:55  <dreimark> ThomasWaldmann: :)
2009-05-28T23:00:57  <dreimark> ThomasWaldmann: dennda we should warn if an acl line is wrong defined on a page
2009-05-28T23:01:54  <dreimark> that should give a similiar message than the one that you were not able to alter it if you don't have admin rights
2009-05-28T23:01:59  * ThomasWaldmann redefines item creation behaviour
2009-05-28T23:04:11  <dreimark> dennda: ThomasWaldmann how do we set a category for any item (not pages)
2009-05-28T23:04:30  <ThomasWaldmann> that will have to go into metadata
2009-05-28T23:04:58  <dreimark> will that be unified for pages than too?
2009-05-28T23:05:08  <ThomasWaldmann> of course
2009-05-28T23:05:13  <dreimark> good
2009-05-28T23:08:14  <ThomasWaldmann> if you visit a non-existing item, it will first ask for mimetype, offering a mimetype selection
2009-05-28T23:08:57  <ThomasWaldmann> after that, it'll show a list of template options for item creation (only showing templates of this mimetype)
2009-05-28T23:09:20  <dreimark> me thinks we should make it to much structured
2009-05-28T23:10:04  <dreimark> we should do it by major mimetypes and guess or examone the correct minor one
2009-05-28T23:10:22  <dreimark> the most of our users are not technically (at least the one I know)
2009-05-28T23:10:44  <dreimark> they think in I have a image, video, text etc
2009-05-28T23:10:59  <dreimark> 23:09 +not
2009-05-28T23:13:39  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5790:775239dc02b4 2.0-storage/MoinMoin/storage/_tests/test_middleware_acl.py: storage: Add initial version if ACLMiddleWare testcases
2009-05-28T23:13:41  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5791:4f730ebdf78f 2.0-storage/MoinMoin/storage/backends/acl.py: storage: first in a series of fixes for the AMW
2009-05-28T23:13:43  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5792:c331f688cbbc 2.0-storage/MoinMoin/storage/_tests/test_middleware_acl.py: storage: add another failing testcase for AMW
2009-05-28T23:13:45  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5793:fbe8c3246354 2.0-storage/MoinMoin/storage/backends/acl.py: storage: replace _backend reference on items that come out of create_item with a reference to the AMW instead of one to the real storage backend. (makes AMW raise NotImplementedError for _commit_item)
2009-05-28T23:13:47  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5794:8f3f50a16d31 2.0-storage/MoinMoin/storage/backends/acl.py:
2009-05-28T23:13:50  <CIA-19> storage: Fix for NotImplementedError thrown by _commit_item. However, this
2009-05-28T23:13:52  <CIA-19> likely requires more references to backend attributes (which isnt too different
2009-05-28T23:14:05  <dreimark> ThomasWaldmann: we could have a slider in user settings for increasing the tech level the user has
2009-05-28T23:14:33  <CIA-19> from what we do when checking permissions and then calling the method of the
2009-05-28T23:14:33  <CIA-19> real backend)
2009-05-28T23:14:33  <CIA-19> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 5795:f196d137e8bb 2.0-storage/MoinMoin/macro/ (4 files in 2 dirs): merge remote 2.0-storage
2009-05-28T23:15:18  <dennda> ThomasWaldmann: can you review those changesets please? it's about how to best fix the "backdoor" introduced by the _backend attribute that's avaliable on items (which, as it was, referenced the real storage backend).
2009-05-28T23:15:33  <dennda> i'd just like to read another opinion as this is a rather important thing
2009-05-28T23:20:25  <ThomasWaldmann> backdoor?
2009-05-28T23:20:36  <dennda> well yeah
2009-05-28T23:21:00  <dennda> if you do foo = amw.get_item("foo")
2009-05-28T23:21:47  <dennda> then foo._backend is a reference to the *real* backend, not the acl wrapper. and this reference is used by all item and revision methods that are usually dispatched to the internal backend classes
2009-05-28T23:21:58  <dennda> it's a bit difficult to get ones head around it
2009-05-28T23:24:34  <dennda> so if you do the above, you can happily call rev = foo.create_revision(..) and that will work without ever checking any acl because it operates on the real storage backend directly
2009-05-28T23:25:28  <dennda> i fixed that by replacing the reference to the real storage backend by a reference to the acl wrapper in the two methods of the acl wrapper where you can retrieve items from
2009-05-28T23:26:54  <dreimark> dimazest: if set or uniq members is a requirement then i think it should check (assert) for that.
2009-05-28T23:26:57  <dimazest> is with moinmo.in wiki everything is ok? when I try to save a page, I get timeouts
2009-05-28T23:28:30  <dimazest> dreimark: in the unittests?
2009-05-28T23:29:05  <ThomasWaldmann> dimazest: works for me
2009-05-28T23:29:50  <dimazest> that's probably my internet, they do some strange things with outgoing traffic
2009-05-28T23:30:13  <ThomasWaldmann> dennda: sounds ok, I'll look at the code later
2009-05-28T23:31:42  <dennda> ok
2009-05-28T23:32:51  <dennda> ThomasWaldmann: the problem is that this of course means that *all* methods defined by the abstract backend must be defined on the acl wrapper as well (at least forwarding to the real storage backend, as I do in my last commit)
2009-05-28T23:40:09  <dimazest> dreimark: tomorrow i'll continue moving GroupDict to the backend. This need changes in other moin code too, like in MoinMoin/events/wikidictsrescan.py see my diary http://moinmo.in/Groups2009/2009-05-28 I need to think on caching api.
2009-05-28T23:40:16  <dimazest> good night
2009-05-28T23:41:26  <dennda> which isn't, as mentioned in the commitmessage, too different from the wrappings we do there anyway...
2009-05-28T23:41:39  <dimazest> i think in MoinMoin/events/wikidictsrescan.py instead of wikidicts.Group GroupManager should be used
2009-05-28T23:42:45  <dimazest> which is taken from request.cfg.group_manager
2009-05-28T23:43:48  <dreimark> sounds good because GroupManager is the replacement for wikidicts.Group
2009-05-28T23:44:34  <dimazest> fine, i'll work on it tomorrow, now i'm too tired
2009-05-28T23:44:44  <dreimark> ok
2009-05-28T23:45:53  <dreimark> not only in the unit test. If it is required to use set in the config backend then there must be a check that it is.
2009-05-28T23:46:31  <dreimark> also if for the meaning of users for acls it is meaned to be uniq and not duplicated users then it need to use set or
2009-05-28T23:46:46  <dreimark> raise
2009-05-28T23:47:20  <dreimark> good night dimazest