MoinMoin: MoinMoinChat/Logs/moin-dev/2009-07-08

2009-07-08T00:18:37  *** xjjk has quit IRC
2009-07-08T00:18:57  * dennda goes zZzzzZzz
2009-07-08T00:18:58  <dennda> n8
2009-07-08T00:27:08  *** starshine_away is now known as starshine
2009-07-08T01:36:56  *** starshine is now known as starshine_away
2009-07-08T02:19:43  *** grzywacz has quit IRC
2009-07-08T07:33:21  <ThomasWaldmann> dennda: "storage: AMW: Why do we accumulate all revisions and then provide a iterator over them instead of yielding them in the first place? Am I missing something obvious or was that just dumb?"
2009-07-08T07:33:50  <ThomasWaldmann> it was because of that: "# TODO: SORT THIS ACCORDINGLY!
2009-07-08T07:35:08  <ThomasWaldmann> but as you might have found out, sorting is not needed, because we already get the items sorted by the wrapped backend and just leaving out some revisions we may not read does change sort order
2009-07-08T08:44:56  <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6018:8da3968b2726 2.0-storage/MoinMoin/storage/backends/ (acl.py newhg.py router.py): storage: comments: fix typos, grammar, style, clarify
2009-07-08T08:46:31  <ThomasWaldmann> dennda: please check your XXX / TODO
2009-07-08T08:48:28  <ThomasWaldmann> 07:35 -> ... does NOT change sort order
2009-07-08T08:54:20  <ThomasWaldmann> dennda: and after that, you just try migrating a 1.8.4 wiki to find the lots of stuff that is still missing
2009-07-08T09:13:00  <ThomasWaldmann> dennda: and to make it clear: you have to document NOW, not "then" (at some undefined time in the future)
2009-07-08T09:16:07  <dreimark> moin
2009-07-08T09:17:24  <ThomasWaldmann> moin dreimark
2009-07-08T09:17:25  <dreimark> ThomasWaldmann: I got yesterday no revisions for every save with the same checksum
2009-07-08T09:17:37  <dreimark> s/no/new/
2009-07-08T09:18:36  <ThomasWaldmann> we don't check what's saved (whether it has changes) currently
2009-07-08T09:18:47  <dreimark> ok
2009-07-08T09:19:24  <ThomasWaldmann> and we currently even sometimes need to save unchanged data (because of changes in the metadata)
2009-07-08T09:24:15  <dreimark> may be we should cache the recent checksum of an item/metadata and verify against that
2009-07-08T09:26:05  <ThomasWaldmann> dennda: read and fix /DEVELOPER_INFORMATION
2009-07-08T12:30:36  <CIA-38> Dmitrijs Milajevs <dimazest@gmail.com> default * 4817:d1996f3b59d5 1.9-groups-dmilajevs/MoinMoin/datastruct/backends/ (__init__.py composite_dicts.py config_dicts.py wiki_dicts.py): Groups2009: retrieve_members renamed to retrieve_items for dict backends. Cosmetic docstring changes.
2009-07-08T13:08:37  <dennda> ThomasWaldmann: How would I best perform that migration? (As there's no fs18 backend...)
2009-07-08T15:28:01  <ThomasWaldmann> dennda: fs17==fs18==fs19
2009-07-08T15:28:37  <dennda> ok
2009-07-08T16:09:21  <dennda> conversion itself went fine
2009-07-08T16:09:56  <dennda> ThomasWaldmann: I'll provide a script for conversion then. What I see missing is mostly UI stuff though
2009-07-08T16:39:17  <dennda> oh and as I don't have a proper 1.8.4 wiki I just used the normal folder structure and copied underlay to pages... (so essentially testing on the same data that the create_persistent_dev_wiki script also provides)
2009-07-08T17:32:10  <dreimark> bbl
2009-07-08T17:56:07  *** amartani has joined #moin-dev
2009-07-08T18:10:33  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6019:2e94b106484d 2.0-storage/MoinMoin/storage/backends/router.py: storage: router: add a comment that we always use normalized item names internally
2009-07-08T18:10:35  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6020:8d39d7f19023 2.0-storage/MoinMoin/storage/backends/ (acl.py newhg.py router.py): merge remote 2.0-storage
2009-07-08T18:10:37  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6021:83fb7769efd2 2.0-storage/ (2 files in 2 dirs): storage: backend cloning script: Allow using the clone function in verbose mode twice
2009-07-08T18:10:39  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6022:9a08941ea2f0 2.0-storage/MoinMoin/storage/backends/enduser.py: storage: In case an incorrect backend_uri is provided, raise a proper error
2009-07-08T18:10:41  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6023:1f405186ef25 2.0-storage/migrate_old_wiki_data.py: storage: First version of migration script for 1.7, 1.8 and 1.9 wikis
2009-07-08T18:10:46  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6024:a1a797edf57d 2.0-storage/create_persistent_dev_wiki.py: storage: dev wiki creation script: update docstring
2009-07-08T18:10:53  <ThomasWaldmann> dennda: we still miss a migration path for existing wikis with existing underlay, including potentially locally modified underlay pages
2009-07-08T18:11:50  <dennda> ThomasWaldmann: ^^^
2009-07-08T18:12:09  <dennda> ThomasWaldmann: With <1.9, if you edit an underlay page, where is that stored? In data/pages or underlay/?
2009-07-08T18:14:12  <dennda> the above does not copy underlay yet. The question is whether we want that (since those potentially old underlay pages will end up in the data backend, indistinguishable from non-underlay pages. we could set a metadata flag or something, though)
2009-07-08T19:00:45  <dennda> http://hg.moinmo.in/moin/2.0-storage/file/a1a797edf57d/MoinMoin/action/GraphInfo.py#l51
2009-07-08T19:00:51  <dennda> what is that comment supposed to tell me?
2009-07-08T19:06:32  <dennda> dreimark: What does PackageItems.py do? I am unsure at the moment whether ACL checks should be performed there or whether that's an administrative task that should not fail due to missing permissions
2009-07-08T19:13:22  <dennda> dreimark: same question for packages.py
2009-07-08T19:13:57  <ThomasWaldmann> dennda: I somehow feel that packages will be done by the xml serialization soon :)
2009-07-08T19:14:36  <ThomasWaldmann> the xml code mostly works already, just the UI / script interface is missing yet
2009-07-08T19:15:33  <ThomasWaldmann> and my point about migration path for underlay was not so much about how to migrate an 1.8 underlay to 2.0, but the general handling of system page upgrades in 2.0+
2009-07-08T19:15:54  <ThomasWaldmann> because we have no underlay any more, stuff will get mixed.
2009-07-08T19:16:05  <ThomasWaldmann> thus:
2009-07-08T19:16:27  <ThomasWaldmann> we need to add 2 metadata entries for "sys pages":
2009-07-08T19:16:37  <ThomasWaldmann> one that is telling "this is a system page"
2009-07-08T19:17:13  <ThomasWaldmann> and another one that is telling "original system page as contained in release <release>"
2009-07-08T19:17:49  <ThomasWaldmann> when someone edits a system page, the "this is a system page" metadata entry will get copied into the edited revision
2009-07-08T19:18:15  <ThomasWaldmann> but the "orig page from release <release>" won't
2009-07-08T19:19:17  <ThomasWaldmann> that way we can now a) what is a system page and what not b) whether someone edited a system page
2009-07-08T19:19:46  <ThomasWaldmann> s/now/know/
2009-07-08T19:21:01  <ThomasWaldmann> then we need support for installing a sys pages package
2009-07-08T19:22:02  <ThomasWaldmann> that would basically deserialize a xml file and create new revisions on top of the existing stuff
2009-07-08T19:22:09  <dennda> could you define what a system page is or give an example?
2009-07-08T19:22:34  <ThomasWaldmann> in that context: everything that lives in underlay in 1.9
2009-07-08T19:22:44  <dennda> ok
2009-07-08T19:23:33  <ThomasWaldmann> we might also need a script that can delete outdated / not-used-any-more sys pages
2009-07-08T19:23:40  <dennda> yes
2009-07-08T19:24:04  <ThomasWaldmann> or list sys pages that miss the "i am original" thing in rev n-1
2009-07-08T19:24:13  <dennda> but since they are rather lightweight (content size) that's not too important for now, i think
2009-07-08T19:24:48  <ThomasWaldmann> it is not for saving space (you remember: we can't nuke items), but rather for cleaning up the namespace
2009-07-08T19:24:56  <dennda> i'm just processing my TODOs
2009-07-08T19:25:11  <dennda> that nuking item thing is another thing
2009-07-08T19:25:16  <dennda> we really need a proper solution
2009-07-08T19:26:29  <ThomasWaldmann> noone is holding you back :D
2009-07-08T19:27:00  <ThomasWaldmann> i guess most backends could do nuke, with the exception of some revision control systems
2009-07-08T19:27:28  <dennda> yes, I'm starting to think that you just shouldn't use any of those for setups where you want that functionality
2009-07-08T19:27:58  <ThomasWaldmann> well, if the API offers nuke, we have to somehow map that
2009-07-08T19:28:07  <dennda> map to what?
2009-07-08T19:28:21  <dennda> to hg and the like?
2009-07-08T19:28:39  <dennda> well they'd just need to ignore the item completely from then on if they can't really delete it
2009-07-08T19:28:47  <ThomasWaldmann> yes, or what should it do when it gets a nuke call via the API?
2009-07-08T19:29:02  <dennda> ignore the item completely
2009-07-08T19:29:15  <dennda> if that's really all we can do
2009-07-08T19:29:22  <ThomasWaldmann> map == emulate
2009-07-08T19:29:45  <ThomasWaldmann> ok, so you see, there is lots to do :)
2009-07-08T19:29:57  <ThomasWaldmann> but i need to move to LUG meeting now :)
2009-07-08T19:30:09  <dennda> most of which not covered by my particular application, by the way :-)
2009-07-08T19:30:43  <dennda> no idea if wikisync.py should go through acl checks, btw
2009-07-08T19:30:45  <ThomasWaldmann> well, if you save the time for underlay, you should help on some replacement
2009-07-08T19:31:01  <ThomasWaldmann> wikisync needs major refactorings anyway
2009-07-08T19:31:15  <dennda> hm?
2009-07-08T19:31:22  <dennda> didn't get your second last statement
2009-07-08T19:31:47  <CIA-38> Dmitrijs Milajevs <dimazest@gmail.com> default * 4818:e8beef3218b9 1.9-groups-dmilajevs/MoinMoin/datastruct/backends/ (wiki_groups.py _formatters.py _tests/test_formatters.py): Groups2009: Formatter is used instead of regex to get group definitions from a page.
2009-07-08T19:32:36  <ThomasWaldmann> System/Help pages (SOCTODO)
2009-07-08T19:32:36  <ThomasWaldmann> Make them work, todo before midterm.
2009-07-08T19:32:37  <ThomasWaldmann> Ideas: layer middleware? just using some metadata flag?
2009-07-08T19:33:09  <ThomasWaldmann> you killed the layering (which is ok for me)
2009-07-08T19:33:23  <ThomasWaldmann> but the metadata stuff is not done yet
2009-07-08T19:34:04  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6025:6880fe93a423 2.0-storage/ (MoinMoin/action/diff.py MoinMoin/action/show.py TODO): storage: ACL: ACL checks are now done on storage layer. Proper UI has to be provided later
2009-07-08T19:34:07  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6026:0e3ab45ef10c 2.0-storage/MoinMoin/action/diff.py: storage: diff should get the item from aclbackend as well so as to not accidentially expose confidential information through diff. AccessDeniedError handling htbdl. Added thoughts
2009-07-08T19:34:09  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6027:5c6b61e2a92c 2.0-storage/MoinMoin/ (5 files in 2 dirs): storage: ACL: ACL checks are now performed on storage layer
2009-07-08T19:34:10  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6028:08b6e21e17c9 2.0-storage/MoinMoin/storage/backends/enduser.py: storage: get_enduser_backend: Automatically create data and user folder
2009-07-08T19:34:22  <dennda> ok, no more TODOs I can fix now
2009-07-08T19:34:29  <dennda> let's go on to XXXs
2009-07-08T19:35:31  <ThomasWaldmann> also, I am not seeing how we should do a beta release. most people won't know what to do with the code due to missing docs.
2009-07-08T19:35:53  <ThomasWaldmann> (admin / user docs I mean in that case)
2009-07-08T19:36:07  <dimazest> ThomasWaldmann: dreimark could you look at the e8beef3218b9 revision. py.test -k TestWikiGroupBackend.test_groups_with_member fails for me and i can't understand why
2009-07-08T19:36:15  <dennda> where should I write those?
2009-07-08T19:36:31  <ThomasWaldmann> in a editor :D
2009-07-08T19:36:31  <dennda> in the wiki I guess. preferences for where exactly?
2009-07-08T19:37:13  <ThomasWaldmann> no
2009-07-08T19:38:03  <ThomasWaldmann> in the end, there needs to be stuff for docs/CHANGES, some config snippets for wiki/config/*
2009-07-08T19:39:14  <dennda> "no, not in the wiki" or "no, i have no preferences"?
2009-07-08T19:39:17  <ThomasWaldmann> also, a developer intro to the storage code is still missing (a verbose description of stuff, maybe in the docstrings, maybe on a wiki page or both)
2009-07-08T19:39:31  <ThomasWaldmann> no pref
2009-07-08T19:43:12  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6029:a30cdec59a0f 2.0-storage/migrate_old_wiki_data.py: storage: PEP8 fixes
2009-07-08T19:43:15  <ThomasWaldmann> gtg/bbl
2009-07-08T20:44:48  *** devilsadvocate has quit IRC
2009-07-08T20:45:01  *** devilsadvocate has joined #moin-dev
2009-07-08T21:15:24  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6030:c185a3868ef6 2.0-storage/MoinMoin/storage/backends/router.py: storage: RouterBackend: Wrap revisions from history
2009-07-08T21:28:22  <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6041:d4caf046bf2a 2.0-storage-editor-amartani/MoinMoin/ (2 files in 2 dirs): Real-time editor: Fetch last revision number periodically, update preview on change.
2009-07-08T21:49:20  <xorAxAx> dennda: why did you ask regarding packagepages? where do you insert the acl checks?
2009-07-08T21:51:14  <dennda> xorAxAx: That code, of which I have no idea what it does, accesses storage. Now I need to know if ACL checks should take place (i.e. using the ACLWrapperBackend that checks and if permission is granted, redirects to the real storage backend) or not (i.e. using the real, unwrapped storage backend without acl checks directly)
2009-07-08T21:51:38  <xorAxAx> of course it needs acl checks
2009-07-08T21:51:43  <xorAxAx> because normal users may use it as well
2009-07-08T21:52:26  <dennda> That's the bit of information I needed
2009-07-08T21:53:04  <dennda> xorAxAx: That is true for both, the code in PackageItems.py and packages.py?
2009-07-08T21:54:12  <xorAxAx> not for packages.py
2009-07-08T21:54:19  <dennda> ok, thanks
2009-07-08T21:56:37  <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6031:67dbfb966ab0 2.0-storage/MoinMoin/action/PackageItems.py: storage: ACLs: PackageItems has to use the AMW since it may be used by regular users as well
2009-07-08T21:59:22  *** grzywacz has joined #moin-dev
2009-07-08T23:47:28  <dreimark> re
2009-07-08T23:47:48  <dreimark> dimazest: me looks at it
2009-07-08T23:54:04  <dreimark> it doesn't fail for me
2009-07-08T23:58:40  <dreimark> currently Empty Page Template shows up with an editor conflict on RC but it hasn't onw