2009-08-05T00:02:42 <ronny> PawelPacana: first i'll figure what i can do about the nasty metadata access in hg tho
2009-08-05T00:04:40 <dennda> ThomasWaldmann: yes it was the strange randomly failing test. still I wonder why all I got is 10 testcases. applying math concludes that it should be less than ten
2009-08-05T00:04:59 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6397:1116feed374a 2.0-storage/MoinMoin/ (3 files in 2 dirs): storage: preload test items in query parsing tests. pull the path to the file up to the common test config
2009-08-05T00:05:01 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6398:d4160e8ec04d 2.0-storage/MoinMoin/_tests/ (test_wikiutil.py wikiconfig.py): storage: merge remote 2.0-storage
2009-08-05T00:12:09 <ronny> PawelPacana: ok, mpm said that implementing wikis on full hg sucks, one should use revlog directly
2009-08-05T00:15:11 <ronny> PawelPacana: the basic issue is, wikis on top of hg wont scale if done in the same way as source trees
2009-08-05T00:15:53 <dennda> Sweet. test_serialization only breaks if issued together with all the other tests. It works if it's run standalone... -_-
2009-08-05T00:17:48 *** tpfennig has joined #moin-dev
2009-08-05T00:20:13 <ronny> TheSheep: did you do any tests on when using hg itself starts to suck too much in hatta
2009-08-05T00:20:33 <ronny> (mpm laid out various scaling issues to me)
2009-08-05T00:20:42 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6399:bce8fcd19ac2 2.0-storage/MoinMoin/ (_tests/test_wsgiapp.py items/__init__.py): fix wsgiapp tests
2009-08-05T00:20:44 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6400:51d54167c47c 2.0-storage/MoinMoin/ (3 files in 2 dirs): merged main
2009-08-05T00:20:45 <ThomasWaldmann> dennda: ^^ read that. and if you hopefully don't expect that we duplicate those 2 long lines everywhere, maybe do what I suggested and compute it once and store it somewhere.
2009-08-05T00:21:15 <dreimark> dennda: packages are only installed by the superuser
2009-08-05T00:22:14 <dennda> ThomasWaldmann: Am I assuming correctly that the serialization tests depend on being run sequentially in the correct order?
2009-08-05T00:22:31 <dennda> hm
2009-08-05T00:22:44 <PawelPacana> ronny: can you point out that issues in detail or provide log?
2009-08-05T00:23:11 <ronny> PawelPacana: mom
2009-08-05T00:23:57 <dennda> ThomasWaldmann: please see 6397:1116feed374a
2009-08-05T00:24:01 <dennda> i did that already.
2009-08-05T00:24:11 <ronny> PawelPacana: http://paste.pocoo.org/show/DfHI3qgMuBg0k7i64DbX/
2009-08-05T00:24:19 <ThomasWaldmann> dennda: let me fix serialization, and fix the num_pages thing instead
2009-08-05T00:24:35 <dennda> ok
2009-08-05T00:25:12 <PawelPacana> ronny: thanks
2009-08-05T00:25:35 <dennda> 8 fails here now
2009-08-05T00:26:18 <ronny> PawelPacana: the basic idea is, small wikis are fine with full hg, the rest doesnt want
2009-08-05T00:26:40 <dennda> "full hg" == mucho data in hg?
2009-08-05T00:27:06 <ronny> full hg == use a real repo and changelogs, have atomic comits
2009-08-05T00:27:18 <ronny> revlogs on themself are very intresting
2009-08-05T00:27:27 <ronny> they expose metadata access
2009-08-05T00:27:32 <ronny> and they are per page
2009-08-05T00:27:39 * dennda is not into that
2009-08-05T00:28:23 <ronny> dennda: there is a breaking point, at wich using revlogs instead of cangelog, maifest and filelogs is better
2009-08-05T00:28:35 <dennda> or 7, if you are lucky with that random test
2009-08-05T00:28:45 <dennda> aha
2009-08-05T00:29:47 <PawelPacana> plain revlog is interesting indeed
2009-08-05T00:38:35 <dennda> Oo
2009-08-05T00:39:10 <dennda> negating a non-existant search term while specifying nothing else will return ALL pages?
2009-08-05T00:39:37 <dennda> shouldn't that rather subtract all hits in the results where that term occurrs?
2009-08-05T00:43:41 *** grzywacz has quit IRC
2009-08-05T00:47:54 <dreimark> test_search needs extremly much time in 2.0-storage
2009-08-05T00:48:10 <dennda> it hung here once
2009-08-05T00:48:12 <dennda> abort and retry
2009-08-05T00:49:11 <dreimark> dimazest: moin-1.9-xapian-dmilajevs/MoinMoin/search/builtin.py:650:52: E231 missing whitespace after ',
2009-08-05T00:49:37 <dennda> I *hate* tests that behave differently depending on how they are run
2009-08-05T00:49:58 <dreimark> looks like some state change
2009-08-05T00:50:11 <dennda> shouldn't happen
2009-08-05T00:50:13 *** amartani has joined #moin-dev
2009-08-05T00:50:14 <dennda> ideally
2009-08-05T00:50:31 <TheSheep> ronny: not really, I don't think about wikis with more than several thousand pages
2009-08-05T00:50:32 * dreimark still waits for test_search
2009-08-05T00:50:47 <TheSheep> ronny: maybe not even seeral hundred
2009-08-05T00:50:48 <dennda> dreimark: seriously. ^C & retry
2009-08-05T00:51:30 <ronny> TheSheep: lets make some tests sometime lates
2009-08-05T00:51:31 <dreimark> now it is faster
2009-08-05T00:52:14 <dennda> don't ask me why it hung
2009-08-05T00:52:15 <dreimark> tests finished: 1838 passed, 7 failed, 91 skipped in 50.44 seconds
2009-08-05T00:52:25 <dreimark> on 2.0-storage using a SSD
2009-08-05T00:52:51 <TheSheep> ronny: I tested it with 2k rfc docs, no noticeable delay
2009-08-05T00:52:55 <dreimark> amartani: can you please merge with 2.0-storage
2009-08-05T00:53:01 <ThomasWaldmann> dennda: if i have separate classes (with or without own Config), shouldn't i get empty backends for each class?
2009-08-05T00:54:01 <dennda> that's the idea
2009-08-05T00:54:33 <ThomasWaldmann> dreimark: the new ones from intel?
2009-08-05T00:54:48 <dennda> actually empty backends for every testcase is what would be correct
2009-08-05T00:54:49 <dreimark> yeah
2009-08-05T00:55:02 <ThomasWaldmann> did you read the news about them?
2009-08-05T00:55:39 <dreimark> not recently
2009-08-05T00:55:41 <dennda> dreimark: I have 47.59 with a 5200 rpm disk
2009-08-05T00:55:57 <ThomasWaldmann> dreimark: read heise/slashdot, iirc yesterday or today
2009-08-05T00:56:29 <dreimark> dennda: the notebook isn't the fastest and it may be depent a bit on the amount of sourcefile checks
2009-08-05T00:56:35 <dreimark> (fresh clone)
2009-08-05T00:59:02 <dennda> hm
2009-08-05T00:59:03 <ThomasWaldmann> dreimark: dirties_backend decorator is still needed?
2009-08-05T00:59:26 <dennda> cls.request = self.parent.request
2009-08-05T00:59:28 <dennda> ?
2009-08-05T00:59:48 <ThomasWaldmann> ehrm, that was for dennda
2009-08-05T00:59:52 <dreimark> known
2009-08-05T01:00:24 <dennda> ThomasWaldmann: not that I'm aware of
2009-08-05T01:00:55 <dennda> but perhaps we should use it
2009-08-05T01:01:17 <dennda> unless you have a better idea as to how we can provide each testcase with a new request object
2009-08-05T01:04:30 <ThomasWaldmann> it doesn't work anyway
2009-08-05T01:04:39 <dennda> hm why?
2009-08-05T01:04:59 <dennda> ah it's still using request.cfg.provide... should be request.provide...
2009-08-05T01:05:24 <ThomasWaldmann> i have separated the xml serialization stuff into separate classes and it can't create a revision because it is already there
2009-08-05T01:05:35 <ThomasWaldmann> thus, it didn't get a clean backend
2009-08-05T01:05:39 <dennda> that is strange
2009-08-05T01:06:08 <dennda> can you push that?
2009-08-05T01:07:53 <dennda> waaah it's past 1 am already
2009-08-05T01:08:49 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6534:c82739c7d57e 2.0-storage-editor-amartani/ (42 files in 24 dirs): merge 2.0-storage
2009-08-05T01:13:33 <ThomasWaldmann> dennda: 8h left :)
2009-08-05T01:20:05 <dennda> pff :/
2009-08-05T01:20:13 <dennda> gn8
2009-08-05T01:21:18 <dreimark> night dennda
2009-08-05T01:24:40 <dreimark> PawelPacana: about the revlog backend. Doesn't have any of these:
2009-08-05T01:24:48 <dreimark> easy data backup distribution (just clone it)
2009-08-05T01:25:09 <dreimark> is clone it the solution ? or is that a comment what not works
2009-08-05T01:26:08 <PawelPacana> sorry, its copied
2009-08-05T01:26:23 <PawelPacana> should be: easy data backup by cloning
2009-08-05T01:27:10 <dreimark> ok
2009-08-05T01:27:58 <dreimark> hmm. me is still not sure if it can clone now or not. because of doesn
2009-08-05T01:28:00 <dreimark> 't
2009-08-05T01:28:56 <dreimark> good night
2009-08-05T01:29:01 <PawelPacana> you should be able to clone now
2009-08-05T01:29:11 <PawelPacana> contents of rev/
2009-08-05T01:29:26 <dreimark> then the doesn't line is confusing
2009-08-05T01:29:40 <PawelPacana> oh, well
2009-08-05T01:29:59 <PawelPacana> you should be able on MercurialBackend
2009-08-05T01:30:18 <PawelPacana> on Revlog not
2009-08-05T01:30:44 <dreimark> ah ok thx
2009-08-05T01:46:50 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6401:62e6ccc6c27d 2.0-storage/MoinMoin/ (conftest.py storage/_tests/test_serialization.py): cleanup conftest, fix serialization tests (and work around non-fresh backend by using different item names in tests
2009-08-05T01:52:34 *** tpfennig has quit IRC
2009-08-05T02:14:12 *** franklin has quit IRC
2009-08-05T02:14:12 *** PawelPacana has quit IRC
2009-08-05T02:14:12 *** dennda has quit IRC
2009-08-05T02:14:12 *** TheSheep has quit IRC
2009-08-05T02:16:02 *** franklin has joined #moin-dev
2009-08-05T02:17:23 *** PawelPacana has joined #moin-dev
2009-08-05T02:17:40 *** dennda has joined #moin-dev
2009-08-05T02:17:55 *** TheSheep has joined #moin-dev
2009-08-05T03:30:57 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6535:bc96f10f3a4e 2.0-storage-editor-amartani/MoinMoin/action/ajax_info.py: Real-time editor: ajax_info transclude item on client request
2009-08-05T03:30:59 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6536:0c2fdab7a726 2.0-storage-editor-amartani/MoinMoin/web/static/htdocs/common/js/text_editor.js: Real-time editor: unified get last revision number and get item transclude into updateInfo, using ajax_info action.
2009-08-05T03:31:01 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6537:00717f4c69c3 2.0-storage-editor-amartani/MoinMoin/action/ajax_info.py: Real-time editor: send transclude if client's revision is not the last.
2009-08-05T03:31:02 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6538:e3a89584dd7d 2.0-storage-editor-amartani/MoinMoin/action/ajax_transclude.py: Real-time editor: Remove ajax_transclude (use ajax_info)
2009-08-05T03:31:04 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6539:eb35bea8058f 2.0-storage-editor-amartani/MoinMoin/action/ajax_info.py: Real-time editor: correct ajax_info description
2009-08-05T03:31:08 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6540:aee5525c0196 2.0-storage-editor-amartani/MoinMoin/web/static/htdocs/common/js/text_editor.js: Real-time editor: send last revision number to ajax_info
2009-08-05T03:31:11 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6541:9ae8f633813b 2.0-storage-editor-amartani/MoinMoin/action/ajax_info.py: Real-time editor: on ajax_info action, fix client revision number evaluation and do lock renew.
2009-08-05T03:31:15 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6542:dae3c20b774e 2.0-storage-editor-amartani/MoinMoin/web/static/htdocs/common/js/text_editor.js: Real-time editor: renew lock using updateInfo.
2009-08-05T05:23:04 *** amartani has quit IRC
2009-08-05T08:24:42 <ThomasWaldmann> moin
2009-08-05T08:54:17 <dennda> moin
2009-08-05T08:54:46 <ThomasWaldmann> moin dennda
2009-08-05T08:55:45 <ThomasWaldmann> see 01:46 changeset comment
2009-08-05T08:56:23 <dennda> already looking at it
2009-08-05T08:56:51 <ThomasWaldmann> if you change foo3 to foo1 in the last test, you'll notice that backends are not fresh
2009-08-05T08:59:17 <ThomasWaldmann> maybe first fix the test_num_pages and the missing script context storage attribute, then you have ~3 less failing tests
2009-08-05T09:00:45 <dennda> ok
2009-08-05T09:20:27 <dennda> ThomasWaldmann: From my understanding of how tests should work in general and with py.test in specific we should find a way to properly provide fresh backends for each and every test case (each method, thus)
2009-08-05T09:21:30 <dennda> because a) in TDD tests should be atomic units independent from each other and b) with py.test, order of tests executed is NOT guaranteed/preserved/forseeable. so there's not much point in relying on it anyway
2009-08-05T09:22:46 <dreimark> dennda: it should be at least optional to use a given wiki instance
2009-08-05T09:23:36 <dreimark> also I think the tests should always behave similiar to a single request
2009-08-05T09:23:42 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6402:ea8db7a6f262 2.0-storage/MoinMoin/parser/_tests/ (test_text_creole.py test_text_moin_wiki.py): fix tests using src code highlighting: they need to use #!highlight ...
2009-08-05T09:24:42 <dennda> dreimark: with wiki instance you mean wiki config?
2009-08-05T09:24:45 <ThomasWaldmann> dennda: that might become rather slow. and not all tests depend on having fresh backends
2009-08-05T09:24:59 <dreimark> dennda: content in a wiki
2009-08-05T09:25:16 <dennda> dreimark: no problem as of late. provide a config with preloaded_xml = ...
2009-08-05T09:27:37 <ThomasWaldmann> btw: 3! :)
2009-08-05T09:27:57 <dennda> nice
2009-08-05T09:28:14 <dennda> are you sure that you covered all pages that the tests need, ThomasWaldmann? (just asking to make sure)
2009-08-05T09:28:55 <ThomasWaldmann> well, at least every page that led to failing tests
2009-08-05T09:29:19 <ThomasWaldmann> and I unified some when it did not matter
2009-08-05T09:30:05 <dennda> ok i will adjust the expected number of results to what we have in the testfile
2009-08-05T09:30:35 <ThomasWaldmann> get it from the clone call
2009-08-05T09:31:09 <ThomasWaldmann> (i mean the total number of test pages)
2009-08-05T09:32:54 <dennda> that happens in wsgiapp and would be rather ugly to return imho. can't i just count the items/revs in storage?
2009-08-05T09:34:31 <dreimark> many failures if you run it with --boxed
2009-08-05T09:34:54 <dreimark> 210 failed
2009-08-05T09:37:32 <dennda> is that the same option we talked about a few weeks ago?
2009-08-05T09:38:24 <ThomasWaldmann> dennda: what is so hard about context.cfg.test_num_pages = clone(...
2009-08-05T09:39:47 <dennda> cloning just to get the amount of revs that *might* have been converted via preloaded_xml?
2009-08-05T09:40:11 <ThomasWaldmann> items, not revs
2009-08-05T09:40:24 <ThomasWaldmann> and you do the clone there anyway
2009-08-05T09:40:26 <dreimark> dennda: yes
2009-08-05T09:40:46 <ThomasWaldmann> in preload_xml
2009-08-05T09:41:01 <dennda> clone returns amount of converted revs (which happens to be the same as the amount of converted items in this case)
2009-08-05T09:41:14 <dennda> dreimark: nothing changed with that
2009-08-05T09:41:27 <ThomasWaldmann> then make it return the amount of items also
2009-08-05T09:42:24 <dennda> just to avoid len(list(storage.iteritems()))? well, ok...
2009-08-05T09:45:46 <ThomasWaldmann> that 'just' saves quite some time. also, you should be aware that list(iteritems()) could fail with fs backend.
2009-08-05T09:46:37 <dennda> for memory backend, which we use to run the tests, it is cheap
2009-08-05T09:47:13 <ThomasWaldmann> not so cheap as using something you almost do anyway
2009-08-05T09:57:03 <ThomasWaldmann> dennda: HelpOnAccessControlLists needs an update
2009-08-05T09:57:53 <dennda> ok
2009-08-05T09:59:24 <ThomasWaldmann> (put it under your project page)
2009-08-05T09:59:38 <dennda> as in move the page?
2009-08-05T10:02:38 <dreimark> moin-1.9 shows for me 899 test run, 68 skipped, 10 failed in 287.98s (rsync: 0.00)
2009-08-05T10:04:03 <dreimark> (boxed)
2009-08-05T10:04:32 <dreimark> moin-2.0-storage 1869 test run, 76 skipped, 210 failed in 62.94s (rsync: 0.00)
2009-08-05T10:11:39 <ThomasWaldmann> dennda: you are working on 2.0 and we don't have a HOACL page for 2.0 yet
2009-08-05T10:12:24 <dennda> ok
2009-08-05T10:13:47 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6403:e73a85eea2d2 2.0-storage/MoinMoin/wsgiapp.py: storage: Fix two search bugs by putting the amount of cloned items into a cfg var
2009-08-05T10:13:49 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6404:9a3e03b2b2be 2.0-storage/MoinMoin/parser/_tests/ (test_text_creole.py test_text_moin_wiki.py): storage: merge remote 2.0-storage
2009-08-05T10:15:08 <dennda> 2 fails here, even with that bloody randomly failing testcase
2009-08-05T10:27:44 <ThomasWaldmann> crap, it only takes 4 minutes here for preload_xml (most for clone to fs storage) to fill the dev wiki
2009-08-05T10:28:13 <dennda> "only"
2009-08-05T10:28:35 <dennda> doesn't take that long here
2009-08-05T10:28:43 <dennda> what system?
2009-08-05T10:31:11 <ThomasWaldmann> TP X300
2009-08-05T10:31:42 <ThomasWaldmann> but the problem is in fs.py :)
2009-08-05T10:32:23 <ThomasWaldmann> for each item you add, it duplicates the cdb with the item name to id mapping
2009-08-05T10:33:05 <ThomasWaldmann> that's a limitation of cdb, it just supports read and create (not update)
2009-08-05T10:33:38 <dennda> uhhh
2009-08-05T10:34:46 <ThomasWaldmann> for normal wiki usage maybe no big problem, but if you want to fill in a lot of pages at once...
2009-08-05T10:35:17 <dennda> it only duplicates the name:id mapping?
2009-08-05T10:35:32 <ThomasWaldmann> yes, record by record :P
2009-08-05T10:36:33 <ThomasWaldmann> maybe we should move clone into storage layer rather than have it on top
2009-08-05T10:37:02 <dennda> what exactly do you mean by that?
2009-08-05T10:37:14 <dennda> urgh, this last failing search test constructs it's private request
2009-08-05T10:37:29 <ThomasWaldmann> e.g. fs.clone would just get a src backend it can read stuff from
2009-08-05T10:38:06 <ThomasWaldmann> then it would just lock, fill the items, build the cdb once, unlock
2009-08-05T10:38:39 <ThomasWaldmann> and be ready after ... 20s? :)
2009-08-05T10:38:43 <dennda> that'd be faster, yes
2009-08-05T10:39:21 <dennda> I would still keep the high level clone
2009-08-05T10:39:56 <ThomasWaldmann> well, there could be a default implementation ...
2009-08-05T10:39:59 <dennda> hm memory backend could be like... def clone(self, other): highlevel_clone(other, self)
2009-08-05T10:40:09 <dennda> yes, that'd be the default implementation
2009-08-05T10:40:11 <dennda> essentially.
2009-08-05T10:40:42 <ThomasWaldmann> you'll reach 0 until noon? :)
2009-08-05T10:41:06 * dennda doesn't get the joke
2009-08-05T10:41:19 <ThomasWaldmann> 0 failures
2009-08-05T10:41:25 <dennda> ah
2009-08-05T10:41:29 <dennda> for you perhaps
2009-08-05T10:41:52 * ThomasWaldmann looks at clone
2009-08-05T10:41:56 <dennda> I still have that strange locking failure from time to time
2009-08-05T10:50:49 <dennda> Oo
2009-08-05T10:51:35 <dennda> localeditlog is b0rked
2009-08-05T10:51:39 <dennda> (stupidly)
2009-08-05T10:53:13 <dennda> that looks like it has never been adapted to the new storage api... *sigh*
2009-08-05T10:55:48 <dennda> that's gonna take a while...
2009-08-05T10:55:49 <dreimark> ThomasWaldmann: fake_mtime makes still trouble
2009-08-05T10:56:54 <dreimark> it is quite close to time.time()
2009-08-05T10:57:32 <dreimark> so it is after the upload time + max_staleness on each request different
2009-08-05T10:59:12 <dennda> Do we want to keep LocalEditLog or is that deprecated?
2009-08-05T11:00:39 <dennda> hm it's only used by Page
2009-08-05T11:00:42 <dennda> afaict
2009-08-05T11:08:18 <ThomasWaldmann> stuff should use rev metadata
2009-08-05T11:08:38 <dennda> page is weird
2009-08-05T11:08:39 <dennda> yes
2009-08-05T11:08:50 <ThomasWaldmann> but just check what is using it before you kill it
2009-08-05T11:08:57 <dennda> only page
2009-08-05T11:09:10 <dennda> is creating an instance of it
2009-08-05T11:09:36 <ThomasWaldmann> check who is using the instance of it :)
2009-08-05T11:10:18 <ThomasWaldmann> dreimark: how can that be? look at how it is computed.
2009-08-05T11:10:43 <dennda> jup
2009-08-05T11:11:22 <dennda> apparently only page itself
2009-08-05T11:18:29 <ThomasWaldmann> xmlrpc theme and Xapian are calling its callers
2009-08-05T11:19:13 <dennda> yes, noticed
2009-08-05T11:19:42 <dreimark> ThomasWaldmann: fake_mtime = int(now / max_staleness) * max_staleness
2009-08-05T11:19:58 <dreimark> lets say max_staleness=10
2009-08-05T11:19:59 <ThomasWaldmann> btw, what are you fixing right now?
2009-08-05T11:20:05 <dennda> me?
2009-08-05T11:20:11 <dennda> that last search bug
2009-08-05T11:20:12 <ThomasWaldmann> yes
2009-08-05T11:20:39 <ThomasWaldmann> you mean script context .storage?
2009-08-05T11:20:43 <dennda> yes
2009-08-05T11:20:47 <ThomasWaldmann> ok
2009-08-05T11:21:17 <ThomasWaldmann> dreimark: yes, and?
2009-08-05T11:22:14 <dreimark> my mistake
2009-08-05T11:22:43 <dreimark> that should work
2009-08-05T11:31:06 <dennda> note: never upgrade to alpha software on machines that you may need
2009-08-05T11:34:39 <dennda> > ImportError: cannot import name AttachFile
2009-08-05T11:34:41 <dennda> %-)
2009-08-05T11:37:34 <ThomasWaldmann> AF is gone, finally
2009-08-05T11:37:55 <dennda> I just deleted everything attachfile related
2009-08-05T11:38:05 <ThomasWaldmann> where?
2009-08-05T11:38:41 <dennda> MoinMoin.search.Xapian._index_page_rev
2009-08-05T11:38:49 <dennda> err
2009-08-05T11:38:58 <dennda> MoinMoin.search.Xapian._index_page
2009-08-05T11:39:54 <dennda> hm now the test works if test_search is run directly, but fails if it's run together with all the other tests... i'd call that progress :-)
2009-08-05T11:44:41 <dennda> btw there's a new profiler for werkzeug powered stuff
2009-08-05T11:44:50 <dennda> > ImportError: cannot import name AttachFile
2009-08-05T11:44:52 <dennda> err
2009-08-05T11:44:54 <dennda> http://screens.ufsoft.org/category/Flickzeug/show/profiler.png
2009-08-05T11:45:58 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6405:8a4a0cf3b4bc 2.0-storage/MoinMoin/action/LocalSiteMap.py: LocalSiteMap action: fix links to use do=
2009-08-05T11:46:00 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6406:057ccbb220c2 2.0-storage/MoinMoin/wsgiapp.py: merge main
2009-08-05T11:46:01 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6407:e43f2bce1eff 2.0-storage/MoinMoin/storage/ (__init__.py backends/__init__.py): move clone/copy_item to the right place (into storage api)
2009-08-05T11:48:24 <ThomasWaldmann> pretty profiler :) does it even work? :D
2009-08-05T11:48:43 <dennda> dunno, ask #pocoo
2009-08-05T11:48:59 <dennda> s0undt3ch is responsible for it
2009-08-05T11:49:02 <dennda> "flickzeug"
2009-08-05T11:50:14 <ThomasWaldmann> btw, that just comes right. i noticed mm2 being rather slow.
2009-08-05T11:52:21 <dennda> me too
2009-08-05T11:52:38 <dennda> sometimes there's some strange loop runnung quite often
2009-08-05T11:52:58 <dennda> i had a print somewhere once which resulted in "EnglishDict" being printed like 100 times
2009-08-05T12:01:15 *** amartani has joined #moin-dev
2009-08-05T12:01:51 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6543:d1ec18d5dc9d 2.0-storage-editor-amartani/MoinMoin/action/ajax_renewlock.py: Real-time editor: Remove ajax_renewlock (use ajax_info)
2009-08-05T12:01:53 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6544:6ddeb4978065 2.0-storage-editor-amartani/MoinMoin/web/static/htdocs/common/js/text_editor.js: Real-time editor: change onload event for jquery.
2009-08-05T12:01:54 <CIA-38> Alexandre Martani <amartani AT gmail DOT com> default * 6545:6418e50f8989 2.0-storage-editor-amartani/MoinMoin/web/static/htdocs/applets/Mobwrite/mobwrite_moinmoin.js: Real-time editor: execute updateInfo with mobwrite.
2009-08-05T12:02:15 <CIA-38> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 6408:9695d66bc11a 2.0-storage/MoinMoin/ (6 files in 5 dirs): clone/copy_item: remove the compat wrappers, call backend methods directly
2009-08-05T12:02:44 <ThomasWaldmann> amartani: moin :)
2009-08-05T12:02:57 <amartani> ThomasWaldmann: moin
2009-08-05T12:07:05 <amartani> just came now to push the last commits, as my connection went down yesterday. going to college now
2009-08-05T12:07:45 <ThomasWaldmann> ah, ok :)
2009-08-05T12:10:28 <dennda> ok Xapian needs major fixing. shall I do that now or skip the test?
2009-08-05T12:11:06 <dreimark> dennda: is there xapwrap involved?
2009-08-05T12:11:19 <dreimark> then it will be changed again after dimazest merge
2009-08-05T12:12:09 <dennda> dreimark: MoinMoin.search.Xapian.Index.{_index_page | _index_page_rev}
2009-08-05T12:13:21 <ThomasWaldmann> yeah, we shouldn't change too much there to avoid lots of merging trouble
2009-08-05T12:13:32 <dennda> ok
2009-08-05T12:15:59 <ThomasWaldmann> dennda: so maybe add that scriptcontext.storage and fix all other new storage code related stuff
2009-08-05T12:16:28 <ThomasWaldmann> if the test is still unhappy then and falls over code that dimazest is working on right now, just add a skip with some appropriate comment
2009-08-05T12:17:15 <dennda> ok
2009-08-05T12:17:19 * dennda discards his local changesets
2009-08-05T12:24:46 <dennda> ok, won't work without refactoring/adjustments to 2.0
2009-08-05T12:24:50 * dennda adds a skip
2009-08-05T12:30:18 <ThomasWaldmann> dennda: now that we are at virtually 0 failing tests, how about one that fails if the test environment isn't working correctly :)
2009-08-05T12:30:42 <dennda> like?
2009-08-05T12:31:13 <ThomasWaldmann> like using preload_xml with the test items and checking if they are there
2009-08-05T12:31:39 <ThomasWaldmann> like not using that and checking if backend is really empty
2009-08-05T12:32:15 <ThomasWaldmann> like having 2 tests that try to create items with same name
2009-08-05T12:32:32 <ThomasWaldmann> test_test_environ.py :)
2009-08-05T12:33:24 <dennda> the latter doesn't make much sense to me
2009-08-05T12:33:27 <ThomasWaldmann> that's also a nice play to show how to use the test environment
2009-08-05T12:33:50 <ThomasWaldmann> it'll fail currently :)
2009-08-05T12:34:21 <ThomasWaldmann> s/play/place/ %)
2009-08-05T12:34:27 <dennda> with 2 tests you mean 2 concurrent editors?
2009-08-05T12:34:41 <ThomasWaldmann> (my brain works phonetically)
2009-08-05T12:34:57 <dennda> oh boy. i fscked my repo again :-)
2009-08-05T12:35:40 <ThomasWaldmann> test_first -> create_item("foo"); write_rev(0)
2009-08-05T12:35:47 <ThomasWaldmann> test_second -> do same
2009-08-05T12:36:12 <dennda> what happens?
2009-08-05T12:36:17 <ThomasWaldmann> second will fail now, because it doesn't get clean backend
2009-08-05T12:36:33 <dennda> ah
2009-08-05T12:36:56 <dennda> in order to fix our test environment. i see
2009-08-05T12:40:33 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6409:f8c9fc0e4818 2.0-storage/MoinMoin/search/_tests/test_search.py: storage: Ladies and gentlemen, we now reached random.randint(0,1) failing tests! Disabled Xapian test due to xapian brokenness
2009-08-05T12:40:35 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6410:c661bc9cb089 2.0-storage/MoinMoin/ (8 files in 7 dirs): storage: merge remote 2.0-storage
2009-08-05T12:40:37 *** tpfennig has joined #moin-dev
2009-08-05T12:41:49 <dennda> funny, working all morning and the thing that came out in the end was a one-liner skip :D
2009-08-05T12:42:55 * dennda updates diary
2009-08-05T12:43:25 <ThomasWaldmann> you can put your insights into the TODO :)
2009-08-05T12:43:32 <dennda> i will
2009-08-05T12:43:43 <ThomasWaldmann> and yes, we should've thought about merge problems first
2009-08-05T12:44:29 <dennda> not that much of a problem. for some reason I enjoyed the entangled code
2009-08-05T12:44:32 <ThomasWaldmann> what's that random failure again?
2009-08-05T12:44:35 <dennda> maybe I should see a therapist
2009-08-05T12:44:47 <dennda> just a sec
2009-08-05T12:45:47 <dennda> http://paste.pocoo.org/show/132645/
2009-08-05T12:46:14 <dennda> I've spoken about that with johill last year
2009-08-05T12:46:25 <xorAxAx> how can that ever work?
2009-08-05T12:46:36 <dennda> iirc it had something to do with some locking granularity depending on some system parameters
2009-08-05T12:46:40 <dennda> (filesystem? no idea, forgot)
2009-08-05T12:46:52 <xorAxAx> it depends on how busy your machine is
2009-08-05T12:47:57 <dennda> aha. can't participate in this discussion
2009-08-05T12:48:18 <xorAxAx> but johill is probably right as well
2009-08-05T12:48:40 <xorAxAx> as you can see, the acquire works for the second time
2009-08-05T12:48:50 <xorAxAx> which is a bit later
2009-08-05T12:52:20 <ThomasWaldmann> lustig, ein test der automatisch skipped, wenn was ned tut :D
2009-08-05T12:54:31 <ThomasWaldmann> dennda: can you try that with different timings?
2009-08-05T12:54:59 <ThomasWaldmann> ext3 = 1s granularity, fat32 = 2s gran.
2009-08-05T12:57:08 <dennda> strange, I changed it but hg diff doesn't show me the change? Oo
2009-08-05T12:57:58 <dennda> ah wrong path %-)
2009-08-05T12:58:11 <dennda> was confused because it worked this time ;)
2009-08-05T12:58:20 <ThomasWaldmann> http://en.wikipedia.org/wiki/Ext4#Improved_timestamps
2009-08-05T12:58:46 <ThomasWaldmann> hopefully that is available via python then also
2009-08-05T12:58:46 <dennda> you mean acquire(2) in both cases?
2009-08-05T12:59:04 <xorAxAx> no, more time
2009-08-05T12:59:31 <ThomasWaldmann> look at teh src :)
2009-08-05T12:59:50 <ThomasWaldmann> and try to not make us wait for too long
2009-08-05T12:59:55 <dennda> timeout is already set to 2.0?
2009-08-05T13:00:11 <dennda> i was under the impression you wanted me to increase it
2009-08-05T13:00:36 <ThomasWaldmann> i want it to run as fast as it can work :)
2009-08-05T13:00:56 <dennda> and I'm only supposed to change timeout?
2009-08-05T13:01:01 * dennda has no understanding of this
2009-08-05T13:01:04 <ThomasWaldmann> look at teh src :)
2009-08-05T13:01:31 <dennda> do you have an irssi macro for that? :P
2009-08-05T13:02:17 <ThomasWaldmann> and it's simple: ext3 ticks like the seconds hand on a clock. it just shows seconds, discrete seconds, 1s granularity.
2009-08-05T13:08:57 <dennda> I've proven empirically (by running the test like 10 times) that 3.0 is ok as timeout for this system under the current load.
2009-08-05T13:09:46 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6411:0c9b09791991 2.0-storage/MoinMoin/util/_tests/test_lock.py: storage: Increase timeout for locking test. Makes it work on my machiene at least.
2009-08-05T13:10:10 <ThomasWaldmann> "machine" :)
2009-08-05T13:10:13 <dennda> shouldn't we increase the minimal timout in ExclusiveLock then as well and clamp every value smaller to that new minimum?
2009-08-05T13:10:54 <ThomasWaldmann> well, I would suggest your first try to explain the conditions that make it work and not just experiment with it
2009-08-05T13:12:22 <dennda> Neither am I a filesystem guy nor is that related to storage.. Do you want me to do that nevertheless? (Just asking with the approaching pencils down date in mind)
2009-08-05T13:15:12 <ThomasWaldmann> well, it doesn't fail for me :)
2009-08-05T13:16:05 <dennda> neither does it fail for me anymore :-)
2009-08-05T13:16:17 <ThomasWaldmann> :P
2009-08-05T13:16:41 <ThomasWaldmann> ok, maybe rather test and fix the test environ
2009-08-05T13:18:17 <ThomasWaldmann> and after that, clean up the rest of "underlay" stuff (just search for that word)
2009-08-05T13:26:37 <ThomasWaldmann> amartani: if you merge moin/2.0-storage now, you should have 0 failing tests.
2009-08-05T13:38:22 <ThomasWaldmann> waldi: ^^
2009-08-05T13:45:07 <dreimark> hmm, now I have 23 tests failing,
2009-08-05T13:48:34 <dreimark> e.g. AttributeError: 'ScriptContext' object has no attribute 'storage'
2009-08-05T13:48:48 <ThomasWaldmann> hg pull -u
2009-08-05T13:49:02 <dreimark> no changes found
2009-08-05T13:49:13 <ThomasWaldmann> hg heads
2009-08-05T13:49:30 <dreimark> changeset: 6411:0c9b09791991
2009-08-05T13:49:33 <dreimark> one head only
2009-08-05T13:50:29 <ThomasWaldmann> hg diff
2009-08-05T13:50:43 <dreimark> nothing
2009-08-05T13:51:18 <ThomasWaldmann> hg up -C
2009-08-05T13:51:46 <dreimark> ok, that changes sthg: 44 files updated, 0 files merged, 3 files removed, 0 files unresolved
2009-08-05T13:52:05 <ThomasWaldmann> then your workdir was on an old revision
2009-08-05T13:52:19 <ThomasWaldmann> your hg version?
2009-08-05T13:52:49 <dreimark> now only 14 failures
2009-08-05T13:54:21 <dreimark> do you have also so much files in /tmp ?
2009-08-05T13:54:44 <dreimark> moin-pOfvKf
2009-08-05T13:55:39 <ThomasWaldmann> do you run the test from the toplevel dir (where wikiserver.py is)?
2009-08-05T13:55:56 <dreimark> yes
2009-08-05T13:56:33 <ThomasWaldmann> hg st - is there anything .py that is not under revision control?
2009-08-05T13:57:14 <dreimark> nothing
2009-08-05T13:58:09 <ThomasWaldmann> test log > pastebin please
2009-08-05T13:59:14 <dreimark> http://moinmo.in/ReimarBauer?action=AttachFile&do=view&target=output.txt
2009-08-05T14:00:18 * dennda looks
2009-08-05T14:01:47 <dennda> is that a fresh clone, dreimark?
2009-08-05T14:02:09 <dreimark> that was a clone from my local repo
2009-08-05T14:02:18 <dennda> hg tip?
2009-08-05T14:02:30 <dreimark> changeset: 6411:0c9b09791991
2009-08-05T14:02:52 <dennda> strange
2009-08-05T14:02:52 <ThomasWaldmann> dreimark: first problem looks like related to sax xml lib
2009-08-05T14:03:10 * dreimark fetches a new clone
2009-08-05T14:03:17 * dennda does that as well
2009-08-05T14:03:29 <ThomasWaldmann> 2nd also
2009-08-05T14:03:50 <ThomasWaldmann> ...
2009-08-05T14:05:35 <ThomasWaldmann> all those 10 are the same error
2009-08-05T14:06:13 <ThomasWaldmann> what python / xml stuff versions is that?
2009-08-05T14:06:22 <dennda> 2.6
2009-08-05T14:08:15 <ThomasWaldmann> i wanted it more precise than visible in the path :)
2009-08-05T14:08:29 * ThomasWaldmann also runs 2.6.2 (ubuntu) and no fails
2009-08-05T14:09:34 <dennda> me to, 2.6.2+
2009-08-05T14:09:40 <dennda> whatever that + means
2009-08-05T14:10:02 * ThomasWaldmann runs tests with py.test --exec=`which python2.5`
2009-08-05T14:12:43 <ThomasWaldmann> hmm, quite some stuff gets a bit unhappy that way
2009-08-05T14:12:53 <dennda> oh noes
2009-08-05T14:13:11 <dennda> but that isn't exactly the same problem as dreimark's, is it?
2009-08-05T14:14:24 <dennda> Can't reproduce with 2.6.2 on fresh clone.
2009-08-05T14:14:31 <ThomasWaldmann> no, but nevertheless interesting :)
2009-08-05T14:14:48 <dennda> sure
2009-08-05T14:14:53 <ThomasWaldmann> maybe suse has fscked up python xml :)
2009-08-05T14:15:17 <dennda> ah dreimark is running suse?
2009-08-05T14:15:26 <ThomasWaldmann> maybe
2009-08-05T14:15:49 <ThomasWaldmann> btw, after changing /usr/bin/python to be python2.5, 0 failures
2009-08-05T14:16:03 <ThomasWaldmann> thus: relax :)
2009-08-05T14:16:09 <dennda> ok so that was a fsckup
2009-08-05T14:16:26 <ThomasWaldmann> maybe some site-packages mixup
2009-08-05T14:16:37 <dennda> let's wait what dreimark gets with the fresh clone
2009-08-05T14:17:36 <ThomasWaldmann> btw, that MoinMoin/_tests/data, is it used?
2009-08-05T14:18:25 <dennda> hm dunno
2009-08-05T14:18:41 <dennda> just added because I was in the middle of something and got sick of adding that manually for every new clone :-)
2009-08-05T14:22:15 <ThomasWaldmann> enduser is still unfinished
2009-08-05T14:22:20 <dennda> yes
2009-08-05T14:22:34 <dennda> i'm unsure about what exactly it should do
2009-08-05T14:22:45 <dennda> perhaps the namespace_mapping should be defined like this:
2009-08-05T14:22:54 <dennda> mountpoint, backend_uri, protection
2009-08-05T14:23:08 <dennda> and then enduser takes care of creating the backends
2009-08-05T14:23:28 <ThomasWaldmann> i think _tests/data is unused
2009-08-05T14:23:43 <dennda> i'm not sure how it should properly "help" to create a namespace mapping (more than providing a sane default)
2009-08-05T14:24:19 <ThomasWaldmann> well, it could provide a sane default for different kinds of backends
2009-08-05T14:24:45 <dennda> indeed
2009-08-05T14:25:12 <dennda> very basic ones only though
2009-08-05T14:25:19 <ThomasWaldmann> btw, it shouldn't be in backends/
2009-08-05T14:25:29 <ThomasWaldmann> rather ..
2009-08-05T14:26:24 <ThomasWaldmann> and it doesn't call router at the end, it just returns the mapping
2009-08-05T14:27:12 <ThomasWaldmann> could be called mapping_creator :)
2009-08-05T14:27:35 <ThomasWaldmann> _helper, _wizard, _whatever
2009-08-05T14:27:56 <dennda> mystical_mapping_master
2009-08-05T14:28:49 <dennda> ThomasWaldmann: s/doesn't/shouldn't/? because it does call router at the moment
2009-08-05T14:30:56 <ThomasWaldmann> i am referring to the sane future implementation :)
2009-08-05T14:31:18 <dennda> what's the benefit?
2009-08-05T14:31:30 <ThomasWaldmann> where you just do mapping = mapping_wizard("fs:/tmp/wiki") and you're done
2009-08-05T14:31:35 <dennda> ah
2009-08-05T14:31:36 <dennda> sure
2009-08-05T14:31:37 <dennda> yeah
2009-08-05T14:32:22 <ThomasWaldmann> depending on $taste, it could even use the old cfg.acl_rights*
2009-08-05T14:32:53 <dennda> elaborate please
2009-08-05T14:33:11 <ThomasWaldmann> but I almost think i would rather like to see them gone and use parameters with defaults
2009-08-05T14:33:22 <ThomasWaldmann> you need acls to create the mapping
2009-08-05T14:33:28 <dennda> yes
2009-08-05T14:33:46 <ThomasWaldmann> for lazy people upgrading their config from 1.9, using the globals would be convenient
2009-08-05T14:34:09 <ThomasWaldmann> but otoh I would rather like to reduce the amount of configuration attributes
2009-08-05T14:34:10 <dennda> they need to adjust that stuff anyway
2009-08-05T14:34:33 <ThomasWaldmann> so better params with sane defaults
2009-08-05T14:35:28 <ThomasWaldmann> and as acl context is not given, you have to use acl_rights_* as param name
2009-08-05T14:36:01 <dennda> jup
2009-08-05T14:36:28 <ThomasWaldmann> hmm, we also need a user acl param if we do not want to hardcode
2009-08-05T14:36:57 <ThomasWaldmann> thus: content_acl_*, user_acl_*
2009-08-05T14:37:13 <ThomasWaldmann> ok, gtg/bbl
2009-08-05T14:53:22 <dreimark> dennda: no kubuntu
2009-08-05T14:54:59 <dreimark> but something is weird still adding file changes
2009-08-05T14:55:25 <dennda> do you still see the test failures on a fresh clone?
2009-08-05T14:55:33 <dreimark> still cloning
2009-08-05T14:55:40 <dennda> ah
2009-08-05T14:55:56 <dreimark> no idea besides that's the FS is quite full
2009-08-05T14:57:42 <dennda> how much space is left?
2009-08-05T14:57:52 <dennda> well, certainly enough if you can still clone
2009-08-05T15:53:11 *** SuBmUnDo_ has quit IRC
2009-08-05T15:57:11 *** SuBmUnDo_ has joined #moin-dev
2009-08-05T16:17:56 <dreimark> dennda: on a fresh clone 14 failures
2009-08-05T16:18:41 <dennda> pff
2009-08-05T16:18:43 <dennda> strange
2009-08-05T16:18:49 <dennda> dreimark: python version?
2009-08-05T16:19:03 <dennda> dreimark: paste the failures please. you only posted 10 iirc
2009-08-05T16:19:19 <dreimark> doing
2009-08-05T16:20:24 <dreimark> http://moinmo.in/ReimarBauer?action=AttachFile&do=view&target=output.txt
2009-08-05T16:28:07 <dreimark> dennda: test_wiki_groups.py[16] ............FFFF
2009-08-05T16:31:36 <dennda> dreimark: those tests do not fail for me either
2009-08-05T16:45:20 <dreimark> strange with boxed I get test_wiki_groups.py[16] ............FF.F
2009-08-05T16:45:40 <dimazest> hmm, let me try
2009-08-05T16:46:16 * dreimark has an idea
2009-08-05T16:47:48 <dimazest> which branch is it?
2009-08-05T16:48:07 <dennda> 2.0-storage
2009-08-05T16:50:51 <dreimark> dennda: http://paste.pocoo.org/show/132689/
2009-08-05T16:50:58 <dreimark> create_page seems to have a problem
2009-08-05T16:51:12 <dreimark> with that patch it is only 1 failing test
2009-08-05T16:51:30 <dreimark> (no idea if that patch makes sense)
2009-08-05T16:52:41 <dennda> py.test docs are so poor...
2009-08-05T16:53:30 <dennda> dreimark: explanation?
2009-08-05T16:54:42 <dimazest> do i need to runt something like py.test --boxed MoinMoin/datastruct/
2009-08-05T16:55:00 <dennda> no --boxed
2009-08-05T16:56:01 <dreimark> dennda: does create_page work if the page exists?
2009-08-05T16:56:08 <dennda> let me check
2009-08-05T16:56:47 <dimazest> 74 passed in 1.33 seconds
2009-08-05T16:57:05 <dreimark> dimazest: hg pull -u ?
2009-08-05T16:57:40 <dreimark> dennda: in the past we have nuked the pages
2009-08-05T16:58:09 <dimazest> pulling from http://hg.moinmo.in/moin/2.0-storage/
2009-08-05T16:58:09 <dimazest> searching for changes
2009-08-05T16:58:11 <dimazest> no changes found
2009-08-05T16:59:27 <dreimark> dennda: does the filesystem type matter
2009-08-05T16:59:43 <dreimark> my home is xfs and it shows these failing tests
2009-08-05T16:59:45 <dennda> it shouldn't
2009-08-05T17:00:04 <dreimark> the /tmp fs is ext3 and it has 10 (-4)
2009-08-05T17:00:12 <dennda> Oo
2009-08-05T17:00:24 <dennda> we don't even use an fs backend for test
2009-08-05T17:01:04 <dennda> What do I need to do in order to execute some code for every test case (every test method, e.g.) in py.test?
2009-08-05T17:04:35 * dreimark son't knwo
2009-08-05T17:04:59 <dreimark> dennda: do you also have lots of tmp files after a test run?
2009-08-05T17:05:42 <dreimark> that could also an explan. because I have a tmp filesystem
2009-08-05T17:07:01 <dennda> yes, there's much stuff
2009-08-05T17:28:20 * dreimark had the same xml failures with py 2.5.4
2009-08-05T17:29:01 <dennda> currently there's some failure introduced by ThomasWaldmann's latest changes
2009-08-05T17:29:10 <dennda> i'm just looking at it
2009-08-05T17:43:02 <dennda> py.test is funny
2009-08-05T18:03:57 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6412:dbc4fa1d6e62 2.0-storage/MoinMoin/storage/__init__.py: storage: Add missing import
2009-08-05T18:04:00 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6413:c9f6e31c8cd6 2.0-storage/MoinMoin/_tests/test_test_environ.py: storage: Add tests that test our test framework
2009-08-05T18:16:05 *** amartani has quit IRC
2009-08-05T18:22:51 *** vinci_ has joined #moin-dev
2009-08-05T18:24:01 <dreimark> dennda: I don't want to tell but now I have 16 failing tests
2009-08-05T18:25:00 <dennda> expected
2009-08-05T18:25:04 <dennda> those two tests should fail
2009-08-05T18:25:17 <dennda> I'm unsure how to fix them correctly
2009-08-05T18:25:40 <dennda> I have an idea that just might work, but if that is fixed that could mean that several different testcases need to be adjusted
2009-08-05T18:39:21 *** tpfennig has quit IRC
2009-08-05T18:51:40 <dreimark> ThomasWaldmann: dennda http://paste.pocoo.org/show/132702/
2009-08-05T18:52:03 <dreimark> this gives no xml failure for my setup
2009-08-05T18:52:15 <dreimark> but has_rights_after
2009-08-05T18:52:19 <dreimark> is None
2009-08-05T18:52:33 <dennda> what is that?
2009-08-05T18:53:11 <dreimark> that is extracted from conftest and the last test of wikiGroups
2009-08-05T18:54:31 <dennda> ThomasWaldmann: Ok after spending hours reading py.test stuff (pointless) and fiddling around I have now come up with a solution that provides fresh backends for each individual test case
2009-08-05T18:55:07 <dennda> ThomasWaldmann: the downside is: there are 34 testcases that rely on the backend being filled with something. do you really want to fix them all now before pencils down?
2009-08-05T18:55:32 <TheSheep> the best py.test documentation is the tests of py.test itself
2009-08-05T18:56:10 <dennda> well I even looked at how django does it out of desparation
2009-08-05T18:56:26 <dennda> and now I'm performing some dark python magic
2009-08-05T18:56:35 <dennda> to trick py.test
2009-08-05T18:56:46 <dennda> (which shouldn't actually matter since that's all dark magic anyway)
2009-08-05T18:57:15 * TheSheep tilts his head
2009-08-05T18:57:18 <TheSheep> really?
2009-08-05T18:57:36 <dennda> well
2009-08-05T18:58:38 <dennda> I've not found a way to change some testcase's (testcase being a method) variables, so I wrote a wrapper around setup_method that does the trick and then invokes the class' setup_method
2009-08-05T18:59:55 <ThomasWaldmann> dennda: what's the problem? you can feed the testpages into them.
2009-08-05T19:00:39 <dennda> it's not always the testcases
2009-08-05T19:00:46 <dennda> s/testcases/testpages/
2009-08-05T19:01:06 <dennda> let me commit what I have here so you can review it before I go on and find myself having done something wrong..
2009-08-05T19:02:02 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6414:792af648e23d 2.0-storage/MoinMoin/ (_tests/test_test_environ.py conftest.py): storage: Apply a trick to provide fresh backends for each and every individual test case. This breaks all testcases that rely on something to be in the backend
2009-08-05T19:02:14 <dennda> hm I should comment that
2009-08-05T19:08:16 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6415:afa6f1f3fc3c 2.0-storage/MoinMoin/conftest.py: storage: add docs for last changeset
2009-08-05T19:08:23 <dennda> review please ^ :-)
2009-08-05T19:08:28 <dennda> TheSheep: you're invited as well
2009-08-05T19:10:54 * TheSheep is at work :(
2009-08-05T19:11:52 <dennda> np
2009-08-05T19:11:56 <dennda> quit
2009-08-05T19:23:36 <dimazest> search tests become too slow....
2009-08-05T19:53:01 <PawelPacana> i wonder why cannot new request be passed to every test which needs it?
2009-08-05T19:54:38 <PawelPacana> since py.test 1.0 there is less hackery needed to provide such fixtures
2009-08-05T19:54:56 <dennda> PawelPacana: tests are not guaranteed to be executed in the order you write them down, and test cases should not depend on each other anyways
2009-08-05T19:55:35 <PawelPacana> i didnt say anything which requires any of that
2009-08-05T19:56:48 <PawelPacana> i just want to say that we dont have to bind request to class and refresh it every test
2009-08-05T19:57:01 <PawelPacana> http://codespeak.net/py/dist/test/funcargs.html
2009-08-05T19:57:15 <dennda> so your critique is about test speed?
2009-08-05T19:58:07 <PawelPacana> maybe i didn't follow the discussusion, bu what exact problem do we have now?
2009-08-05T19:58:08 <ThomasWaldmann> maybe we shouldn't use too new py.test features. i like it when it works with dist packages.
2009-08-05T19:59:11 <PawelPacana> s/follow/follow carefully/
2009-08-05T20:02:51 <dreimark> can one solve that miracle
2009-08-05T20:02:54 <dreimark> http://paste.pocoo.org/show/132715/
2009-08-05T20:03:11 <dreimark> name it debug_foo.py and run it in moin-2.0-storage
2009-08-05T20:03:16 <dreimark> my output is
2009-08-05T20:03:21 <dreimark> * ExampleUser
2009-08-05T20:03:21 <dreimark> * AnotherUser
2009-08-05T20:03:21 <dreimark> True
2009-08-05T20:03:21 <dreimark> None
2009-08-05T20:03:21 <dennda> tbh, I didn't see *that* much of a problem (all our tests passed), but from a software technicians point of view it might be cleaner to provide fresh backends for every testcase. Although I have to admit that I start sharing your opinion, PawelPacana
2009-08-05T20:07:26 <PawelPacana> dist packages mostly suck when it comes to newer versions
2009-08-05T20:07:27 <dreimark> and if I enter those commands to a python shell then sometimes has_rights_after becomes True
2009-08-05T20:07:54 <dreimark> for me it looks like acl or groups aren't working
2009-08-05T20:08:29 <dennda> I did not change security.AccessControlList afair
2009-08-05T20:10:42 <dreimark> and the most annoying thing is that it always works if I execute it in /tmp filesystem clone
2009-08-05T20:11:37 * dennda has *no* idea
2009-08-05T20:13:22 <PawelPacana> dreimark: my output is the same
2009-08-05T20:13:48 <dreimark> do you have also a filesystem for /tmp ?
2009-08-05T20:15:43 <dennda> PawelPacana: if you back out my latest py.test-patchery related commits and run the tests, how many failures do you get?
2009-08-05T20:17:32 <PawelPacana> dreimark: ?
2009-08-05T20:18:36 <dreimark> PawelPacana: my home fs is XFS and I have a separate FS for /tmp of ext3
2009-08-05T20:18:51 <PawelPacana> oh, my is ext4
2009-08-05T20:19:10 <PawelPacana> for both
2009-08-05T20:19:13 <dreimark> some hours ago I created a clone in /tmp and wondered why I have 4 tests not failing
2009-08-05T20:19:27 <dreimark> dennda: do you have a separate FS for /tmp ?
2009-08-05T20:19:40 <dennda> no
2009-08-05T20:19:44 <PawelPacana> dreimark: i can remount /tmp to tmpfs
2009-08-05T20:19:56 <PawelPacana> and try again
2009-08-05T20:19:59 <dreimark> ok
2009-08-05T20:20:49 <PawelPacana> dennda: on 6413:c9f6e31c8cd6 6 failed, 1802 passed, 89 skipped, 12 error in 49.73 seconds
2009-08-05T20:21:13 <dennda> ThomasWaldmann: So what do we do now with respect to tests (given the above discussion). I also think that we could still fix those tests after the more important parts of my application are done
2009-08-05T20:21:24 <dreimark> PawelPacana: many of them sax xml failures?
2009-08-05T20:21:52 <PawelPacana> actually i ran with no tracebacks, let me check again
2009-08-05T20:21:58 <dreimark> simiilar to http://moinmo.in/ReimarBauer?action=AttachFile&do=view&target=output.txt
2009-08-05T20:22:11 <dennda> looks like
2009-08-05T20:22:18 <dennda> 10 xml errors + 2 due to the new test cases
2009-08-05T20:25:59 <PawelPacana> http://paste.pocoo.org/show/132720/ on 6413:c9f6e31c8cd6
2009-08-05T20:26:52 * ThomasWaldmann is busy with hot pizza
2009-08-05T20:27:19 <dennda> hehe
2009-08-05T20:27:48 <dennda> PawelPacana: ok great, thanks. that's stuff for our pizza guy :-)
2009-08-05T20:28:38 <PawelPacana> bbl
2009-08-05T20:35:19 <dreimark> also it is funny that me can get MoinMoin/_tests/test_wsgiapp.py[0]
2009-08-05T20:35:42 <dreimark> and PawelPacana sees two tests there
2009-08-05T20:36:18 <dreimark> ThomasWaldmann: guten appetit
2009-08-05T20:40:56 <ThomasWaldmann> jo, fertig :)
2009-08-05T20:55:25 <dennda> Oh and perhaps we should have some roadmap
2009-08-05T21:12:54 <dreimark> ThomasWaldmann: dennda http://paste.pocoo.org/show/132726/
2009-08-05T21:13:07 <dreimark> that solves my xml trouble
2009-08-05T21:13:30 <dennda> aha
2009-08-05T21:17:24 <PawelPacana> dreimark: on tmpfs /tmp test results are the same as before
2009-08-05T21:19:15 <dreimark> hmm
2009-08-05T21:19:34 <dreimark> with that patch I have http://moinmo.in/ReimarBauer?action=AttachFile&do=view&target=output.txt only 6 failing tests now
2009-08-05T21:20:29 <dennda> dreimark: hg tip?
2009-08-05T21:23:11 <dreimark> no sorry have reverted to hg revert -r 6413 --all
2009-08-05T21:23:54 <ThomasWaldmann> dreimark: ah, that's a known fsckup
2009-08-05T21:24:04 <dennda> it is?
2009-08-05T21:24:15 <dreimark> can I commit that or do you have a better idea?
2009-08-05T21:24:43 <dennda> i wonder how you found that
2009-08-05T21:25:15 <dreimark> by investigating this http://paste.pocoo.org/show/132728/
2009-08-05T21:25:40 <dreimark> and btw. we need to know how much tests we in total have
2009-08-05T21:25:48 <ThomasWaldmann> dennda: i remember having had that "in does not work, has_key works" somewhere else in xml stuff also
2009-08-05T21:26:03 <dreimark> because I would never got that idea if I haven't seen that PawelPacana has more tests than me
2009-08-05T21:26:05 <dennda> aha
2009-08-05T21:26:12 <dennda> what a mean bug
2009-08-05T21:26:56 <dreimark> ThomasWaldmann: can you look at the last pastebin as debug_blub.py in moin-2.0-storage why it does not find the page object?
2009-08-05T21:30:09 <ThomasWaldmann> dreimark: ?
2009-08-05T21:31:52 <dreimark> I gert this result http://paste.pocoo.org/show/132729/ for running http://paste.pocoo.org/show/132728/
2009-08-05T21:32:13 <dreimark> AttributeError: Formatter instance has no attribute 'page'
2009-08-05T21:33:41 <ThomasWaldmann> yeah, that stuff seems a bit messed up
2009-08-05T21:33:48 <ThomasWaldmann> but page is deprecated anyway
2009-08-05T21:34:17 <dreimark> this is from test_wsgiapp
2009-08-05T21:34:35 <dreimark> if it can't work we better skip that test
2009-08-05T21:40:07 <CIA-38> Reimar Bauer <rb.proj AT googlemail DOT com> default * 6416:a91db47c2e1b 2.0-storage/MoinMoin/storage/serialization.py: storage.serialization: we need to test by "has_key" at an instance of AttributesImpl (attrs)
2009-08-05T21:43:36 <dreimark> dennda: tmp FS clone 33 failed and on home FS clone 37 now
2009-08-05T21:43:55 <dennda> hm?
2009-08-05T21:44:05 <dreimark> with tip
2009-08-05T21:45:35 <dennda> well yes. dunno if we want to keep it that way or just revert the py.test patch and fix that after the more essential stuff is done (/me looks at ThomasWaldmann )
2009-08-05T21:54:17 * dreimark looks at appendpage
2009-08-05T21:54:39 <CIA-38> Dmitrijs Milajevs <dimazest@gmail.com> default * 4935:2171281b6d79 1.9-xapian-dmilajevs/MoinMoin/search/_tests/test_search.py: Xapian2009: Search tests update. regex: and case: prefixes tests were added for title, linkto and category.
2009-08-05T22:01:36 *** grzywacz has joined #moin-dev
2009-08-05T22:07:03 <dimazest> dreimark: could you review the test. it is a little bit tricky, may be there is a better way to do it. I still need to write test for language: mimetype: and domain:
2009-08-05T22:07:48 <dimazest> are regex and case are supposed to be used with them? (e.g. domain:case:somedomain)
2009-08-05T22:08:09 <dimazest> and the unicode tests are still in my todo
2009-08-05T22:08:45 <dimazest> btw, these tests are quite slow
2009-08-05T22:08:56 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6417:644aec6b0c26 2.0-storage/ (MoinMoin/Page.py MoinMoin/web/contexts.py docs/INSTALL.wsgi): storage: remove underlay leftover
2009-08-05T22:09:01 <CIA-38> Christopher Denter <moin GUESSWHAT the DASH space DASH station ROUNDTHING com> default * 6418:43645d8dc00f 2.0-storage/MoinMoin/storage/serialization.py: storage: merge remote 2.0-storage
2009-08-05T22:20:20 <dreimark> dimazest: I will look, need currently some time to understand some item issue
2009-08-05T22:21:53 <dreimark> for mimetype i probably one likes to add a regex e.g. for all text types
2009-08-05T22:22:13 <dimazest> ok, no problem for me, ich muss Deutschaufgabe machen, wie gestern
2009-08-05T22:22:18 <dreimark> for language I think rather not
2009-08-05T22:59:25 <dreimark> damn fsck. after that ntfs issue I should have got that earlier
2009-08-05T23:00:14 <dreimark> cache_mtime is float while page_mtime is integer
2009-08-05T23:02:09 <dennda> hm looks like thomas fell right asleep while digesting his delicious meal :D
2009-08-05T23:04:19 <dreimark> dimazest: http://hg.moinmo.in/moin/1.9/file/7a8e1567e7e4/MoinMoin/datastruct/backends/wiki_groups.py#l43
2009-08-05T23:04:35 <dreimark> the comparision needs to better done on the lowest type of both
2009-08-05T23:04:54 <dreimark> if one of them is integer it should be done in integer
2009-08-05T23:07:12 <dreimark> compatision on integer fixes my FS issue with wiki_groups tests in 2.0-storage
2009-08-05T23:11:45 <dreimark> dennda: http://paste.pocoo.org/show/132742/ I think that is better you may want to rename it to append_item or something similiar
2009-08-05T23:12:27 <dreimark> I am not sure abot the mimetype (if it is allowed to change it or where it is)
2009-08-05T23:29:00 <dreimark> dimazest: I am not sure if I get currently what the meaning of case is
2009-08-05T23:37:24 <dreimark> dimazest: please refactor the other tests in test_search also to new style classes
2009-08-05T23:40:26 <dimazest> ok
2009-08-05T23:40:48 <dimazest> dreimark: should i change wiki_groups.py?
2009-08-05T23:40:56 <dreimark> yes
2009-08-05T23:41:11 <dreimark> and add a comment
2009-08-05T23:41:40 <dreimark> (it is better we read that often)
2009-08-05T23:43:55 <dreimark> dimazest: if the amount of hits is known it would be also good to assert for that
2009-08-05T23:45:20 <dreimark> in TestSearch
2009-08-05T23:46:54 <dreimark> the test don't currently use xapian search or?
2009-08-05T23:47:18 * dreimark needs to get some sleep
2009-08-05T23:47:21 <dreimark> good night
2009-08-05T23:47:32 <PawelPacana> gn MoinMoin: MoinMoinChat/Logs/moin-dev/2009-08-05 (last edited 2009-08-04 22:15:02 by IrcLogImporter)
