MoinMoin: MoinMoinChat/Logs/moin-dev/2006-09-02

2006-09-02T02:21:28  <xorAxAx> ThomasWaldmann: the mentors filled out their survey, right?
2006-09-02T02:25:45  <xorAxAx> esp. mine
2006-09-02T02:26:53  <ThomasWaldmann> yours is submitted
2006-09-02T02:27:04  <ThomasWaldmann> when is the deadline again?
2006-09-02T02:28:06  <xorAxAx> in 6 days
2006-09-02T03:10:24  * ThomasWaldmann notified richard to do it soon
2006-09-02T03:33:38  <xorAxAx> antispam = 49.286s -- hmm
2006-09-02T09:42:08  <ThomasWaldmann> moin
2006-09-02T11:26:23  <Kepplar> moin
2006-09-02T12:16:06  <xorAxAx> Kepplar: have you tackled the diagrams already? :)
2006-09-02T14:54:59  <xorAxAx> ThomasWaldmann: can we shutdown the sync and xapian branches, merge CHANGES and give main commit permissions to franz?
2006-09-02T14:55:16  <xorAxAx> shutdown as in - disallow commits
2006-09-02T14:55:20  <xorAxAx> google might want to look at them
2006-09-02T15:08:16  <ThomasWaldmann> sure
2006-09-02T15:11:13  <ThomasWaldmann> fpletz: xapian branch is ro now, but you have rw access to main branch
2006-09-02T15:11:26  <ThomasWaldmann> xorAxAx: sync branch ------""-------
2006-09-02T15:11:36  <xorAxAx> ok
2006-09-02T15:11:44  <ThomasWaldmann> Kepplar: your branch is still rw
2006-09-02T22:52:26  <ThomasWaldmann> xorAxAx: http://moinmoin.wikiwikiweb.de/MoinMoinTodo/Release_2.0/StepByStep
2006-09-02T22:52:33  <ThomasWaldmann> see latest edits there
2006-09-02T22:53:20  <xorAxAx> e.g. "make user data stuff easier by handling it by existing code and the same way as pages (might degrade performance temporarily, but makes modularization easier):"
2006-09-02T22:53:33  <xorAxAx> is a little bit hard if there isnt a kind of meta api yet
2006-09-02T22:54:16  <xorAxAx> more concretely said: changing the storage format doesnt help at all if there is not stable api ...
2006-09-02T22:54:37  <xorAxAx> so there are a few steps missing
2006-09-02T22:55:26  <xorAxAx> or maybe this "modularize" step is not ... stressed enough :=)
2006-09-02T22:55:37  <xorAxAx> just moving methods is not enough
2006-09-02T22:56:56  <xorAxAx> ThomasWaldmann: btw, does your 1.4 code use a different storage format?
2006-09-02T22:59:27  <ThomasWaldmann> http://hg.thinkmo.de/moin/2.0?f=58437df10b1c;file=MoinMoin/storage.py == yes
2006-09-02T22:59:41  <ThomasWaldmann> but not in the way I think it is best now
2006-09-02T23:01:34  <xorAxAx> so i think the next steps are your "remove ideas" and then a harsh refactoring to insert a storage class while retaining some basic amounts of the Page interface (but while not doing any specific changes to the storage format)
2006-09-02T23:02:01  <xorAxAx> and i am sure that this can be done in less than 3 months "part time" :)
2006-09-02T23:04:02  <ThomasWaldmann> we also need a user api
2006-09-02T23:04:11  <ThomasWaldmann> (storage api)
2006-09-02T23:04:27  <ThomasWaldmann> so first unifying it would be less work
2006-09-02T23:05:04  <xorAxAx> your idea of user data being normal items etc. and then subclassing ... in that case you need some base classes first :)
2006-09-02T23:05:15  <xorAxAx> but you suggest to change the storage formt
2006-09-02T23:05:19  <xorAxAx> that is not necessary
2006-09-02T23:05:49  <xorAxAx> you can expose userdata that resides in the old format as items
2006-09-02T23:07:21  <ThomasWaldmann> the point is what will make more trouble. keeping old storage formats or changing it.
2006-09-02T23:07:47  <xorAxAx> my point is that if keeping the old format makes troubles, then the API wasnt well designed :)
2006-09-02T23:08:05  <xorAxAx> and the aforementioned user thing isnt a problem, i can implement it if anybody doubts it :)
2006-09-02T23:08:20  <xorAxAx> but i see a different problem that kepplar didnt solve either
2006-09-02T23:08:30  <xorAxAx> what to do with the combined rev logs?
2006-09-02T23:08:35  <xorAxAx> (aka. global edit-log)
2006-09-02T23:09:16  <ThomasWaldmann> that has to be seen at the very end :)
2006-09-02T23:09:30  <xorAxAx> ?
2006-09-02T23:09:31  <ThomasWaldmann> we maybe could get rid of that duplication
2006-09-02T23:09:32  <xorAxAx> %-)
2006-09-02T23:09:44  <ThomasWaldmann> by using local edit-log and caching
2006-09-02T23:09:55  <xorAxAx> i am just talking about the API
2006-09-02T23:09:59  <xorAxAx> not about the implementation
2006-09-02T23:11:29  <xorAxAx> one could e.g. query the last x revisions without filtring for a special name
2006-09-02T23:11:37  <xorAxAx> then you have your global edit-log :)
2006-09-02T23:11:38  <ThomasWaldmann> maybe we could have a new api there
2006-09-02T23:11:50  <xorAxAx> the whole thing should be "new"
2006-09-02T23:12:01  <xorAxAx> i dont want to see anything old there, besides legacy "delegators"
2006-09-02T23:12:12  <xorAxAx> (that arent a problem either)
2006-09-02T23:14:13  <xorAxAx> has kepplar documented anything in the last months about his design?
2006-09-02T23:14:42  <xorAxAx> i remember the wiki page to have remained pretty unchanged
2006-09-02T23:15:04  <Kepplar> i have design docs on my desktop
2006-09-02T23:15:16  <Kepplar> (sorry i will be back tomorrow ive been working my socks off today on non-moin stuff)
2006-09-02T23:17:31  <xorAxAx> so writing some abstract classes, and another implementation subclass that maps to current storage code should be the next (after "remove this and that") goal
2006-09-02T23:19:40  <xorAxAx> ThomasWaldmann: what do you think?
2006-09-02T23:20:59  <ThomasWaldmann> cache and attachment stuff separation should be also done before
2006-09-02T23:21:36  <xorAxAx> whats to be done there?
2006-09-02T23:21:56  <xorAxAx> you nicely extended caching.py to get some attributes that delegate the place of storage
2006-09-02T23:22:09  <ThomasWaldmann> making it independand of page object
2006-09-02T23:22:31  <xorAxAx> in the new storage API, attachments will be normal items
2006-09-02T23:22:48  <xorAxAx> so the storage code will simply have to search them in the attachment folder in some cases
2006-09-02T23:23:22  <xorAxAx> this "making independent" step would be automatically included when writing the implementation code for the new API
2006-09-02T23:24:25  <xorAxAx> and when and where it searches is defined by some migration semantics that have to be thought about in any case
2006-09-02T23:25:03  <xorAxAx> (that answers questions like "what if there is already a subpage called foo.txt but the parent page has an attachment with that name?)
2006-09-02T23:25:18  <ThomasWaldmann> maybe just add your ideas as more small steps to that page
2006-09-02T23:25:24  <xorAxAx> ok
2006-09-02T23:25:39  <xorAxAx> oh, the browser has been saving for a few minutes
2006-09-02T23:25:47  <xorAxAx> 12 minutes
2006-09-02T23:32:10  * ThomasWaldmann sees double
2006-09-02T23:34:04  <xorAxAx> me too
2006-09-02T23:34:14  * xorAxAx is saving
2006-09-02T23:34:56  <xorAxAx> hmm, still saving
2006-09-02T23:35:00  <xorAxAx> one minutre
2006-09-02T23:35:24  <xorAxAx> antispam = 0.531s compile_huge_and_ugly = 0.632s getACL = 0.516s load_multi_cfg = 0.000s parsePagelinks = 0.083s run = 74.569s send_page = 1.248s send_page_content = 0.846s total = 77.771s
2006-09-02T23:35:35  <xorAxAx> hmm, we need a timer around the mailing code
2006-09-02T23:45:19  <ThomasWaldmann> feel free...
2006-09-02T23:45:53  <xorAxAx> sure
2006-09-02T23:45:54  * ThomasWaldmann is still thinking about how to get that _ ambiguity out again
2006-09-02T23:46:11  <xorAxAx> hehe
2006-09-02T23:46:33  * xorAxAx is currently setting up the sync setup for the notepad wiki
2006-09-02T23:49:20  <ThomasWaldmann> esp. about how to do it in a mostly automatic way