MoinMoin: MoinMoinChat/Logs/moin-dev/2008-08-05

2008-08-05T00:39:50  <CIA-53> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 4277:b30d0384b136 1.8-dom-bblank/MoinMoin/ (macro2/NewPage.py macro/NewPage.py): Migrate NewPage macro to new interface
2008-08-05T01:50:10  <ThomasWaldmann> byeongweon: if you close a bug, don't forget to change the category at the end of the page
2008-08-05T01:50:56  <byeongweon> ThomasWaldmann: oh. I didn't know that.
2008-08-05T01:51:09  <byeongweon> thanks for notify
2008-08-05T01:52:13  <byeongweon> you mean add CategoryMoinMoinBugFixed category. right?
2008-08-05T01:54:47  <ThomasWaldmann> no, don't add, change
2008-08-05T01:55:58  <byeongweon> ThomasWaldmann: ok. I understand
2008-08-05T02:29:52  <CIA-53> Byeongweon [tasyblue@gmail.com] default * 3961:bdb71fda36eb 1.8-guieditor-mbyeongweon/MoinMoin/ (7 files in 3 dirs): merge with moin-1.8 main branch
2008-08-05T02:29:56  <CIA-53> Byeongweon [tasyblue@gmail.com] default * 3978:529b71b82fdc 1.8-guieditor-mbyeongweon/ (14 files in 9 dirs): merge with moin 1.8 main branch
2008-08-05T02:29:57  <CIA-53> Byeongweon [tasyblue@gmail.com] default * 3979:40f10410e0b3 1.8-guieditor-mbyeongweon/MoinMoin/ (2 files in 2 dirs): fix bug : MoinMoinBugs/GuiEditorDestroysParserSyntax
2008-08-05T08:46:37  <ThomasWaldmann> moin
2008-08-05T08:48:09  <ThomasWaldmann> byeongweon: http://moinmo.in/4ct10n/diff/MoinMoinBugs/GuiEditorDestroysParserSyntax?action=diff&date=1217893895000000
2008-08-05T08:48:57  <ThomasWaldmann> you removed the word "Category", this is of course incorrect
2008-08-05T08:56:39  <dreimark> moin
2008-08-05T09:01:42  <ThomasWaldmann> hi dreimark
2008-08-05T09:03:14  <ThomasWaldmann> byeongweon: http://moinmo.in/MoinMoinBugs/TableCellNewLine new bug?!
2008-08-05T09:21:23  <dreimark> that one is not new and in text mode it can't work also
2008-08-05T09:21:41  <dreimark> it is doing something like ||a\n\n||
2008-08-05T09:22:14  <dreimark> it can only work if one replaces \n by <<BR>> and I think this was discussed by xorAxAx and byeongweon a few days ago
2008-08-05T09:23:05  * dreimark reboots bbl
2008-08-05T09:34:12  <ThomasWaldmann> gizmo__: 5 days no commit :|
2008-08-05T10:45:46  <dreimark> byeongweon: i have problems with getting fck started in ff 2.0.0.16
2008-08-05T10:46:30  <dreimark> I don't get the icons
2008-08-05T11:05:48  <CIA-53> Bastian Blank <bblank@thinkmo.de> default * 4278:8a392d3dab34 1.8-dom-bblank/MoinMoin/converter2/ (link.py _tests/test_link.py): Link converter - Handle query strings and anchor more correctly, add testcases
2008-08-05T11:06:02  <waldi> this needs a real uri parser ...
2008-08-05T11:28:04  <waldi> and urlparse does not allow configuration of new url schemes, which makes the parsing a little bit flacky
2008-08-05T11:57:51  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4442:96480deedb7a 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/research/fileheads.py: (hg) Small hg extension to show file headsfrom console. Mostly for debugging.
2008-08-05T11:59:00  <zenhase> hooray! now i have a proper suspensory bandage :)
2008-08-05T12:36:55  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4443:e2a83e8d8f2b 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/research/showextra.py: (hg) Hg extension to show extra dict from console.
2008-08-05T15:21:14  <dennda> slow cia
2008-08-05T15:23:15  * johill wonders if dennda ever runs fs tests
2008-08-05T15:25:30  <dennda> yes, I do. Why?
2008-08-05T15:25:40  <johill> two fail
2008-08-05T15:25:55  <dennda> I know
2008-08-05T15:26:22  <dennda> I can't say that much about the FS backend though, since you were the one who wrote it
2008-08-05T15:26:55  <dennda> (ok, my own backend has two failing tests as well)
2008-08-05T15:26:56  <johill> well it's just failing in the test that you wrote the test correctly
2008-08-05T15:26:56  <johill> :)
2008-08-05T15:27:14  <johill> pushed a fix
2008-08-05T15:27:29  <dennda> that sentence makes little sense to me :)
2008-08-05T15:27:55  <johill> it fails in check(are tests written properly)
2008-08-05T15:28:05  <dennda> you will push a fix, when CIA is ready
2008-08-05T15:28:23  <johill> %^$#%^@#
2008-08-05T15:28:25  <dennda> I am pushing for ten minutes now
2008-08-05T15:28:37  <johill> your merge is there
2008-08-05T15:28:40  <johill> so you can abort now
2008-08-05T15:28:44  <johill> and if you abort, I can push :)
2008-08-05T15:28:53  <dennda> yes, the other one too
2008-08-05T15:29:08  <johill> other one?
2008-08-05T15:29:10  <dennda> does it only accept one connection at a time!?
2008-08-05T15:29:18  <johill> just check whether your last changeset is up on the server
2008-08-05T15:29:19  <dennda> storage: action/AttachFile.py: Make _do_get basically work.
2008-08-05T15:29:21  <johill> and if it is, abort
2008-08-05T15:29:32  <johill> that's there
2008-08-05T15:29:40  <dennda> remote: Killed by signal 2.
2008-08-05T15:29:44  <johill> I think hg locks the repo
2008-08-05T15:29:44  <dennda> there you go
2008-08-05T15:29:49  <dennda> ah
2008-08-05T15:30:01  <dennda> not so great
2008-08-05T15:31:44  * dennda slaps CIA-53 with a large hydron collider
2008-08-05T15:32:04  <ThomasWaldmann> hehe
2008-08-05T15:32:12  <ThomasWaldmann> dennda: #cia
2008-08-05T15:32:31  <dennda> Yes, I know
2008-08-05T15:33:48  <johill> ThomasWaldmann: it doesn't matter, if the tools continually get in the way it's not interesting whether the problem is known or not
2008-08-05T15:34:53  <dennda> I wonder if that problem is caused by software or hardware issues
2008-08-05T15:34:55  <johill> ThomasWaldmann: we still blame you, because by now it's well established that cia just doesn't work properly
2008-08-05T15:36:30  <dennda> Isn't it possible to make that thing not lock up the whole push-process?
2008-08-05T15:36:49  <dennda> I mean, one could periodically check if there are new changes and just send out a message to the channel
2008-08-05T15:40:23  <johill> now hg is confused
2008-08-05T15:40:26  <johill> my push isn't working at all
2008-08-05T15:40:41  <johill> just hangs at "searching for changes"
2008-08-05T15:41:02  <dennda> sweet
2008-08-05T15:41:26  <dennda> let's do tarballs and patchfiles
2008-08-05T15:42:24  * dennda hugs CIA-53 
2008-08-05T15:42:25  * CIA-53 hugs dennda
2008-08-05T15:42:30  <johill> lol
2008-08-05T15:42:35  <dennda> Well, it seems there is a bottleneck
2008-08-05T15:42:55  <dennda> since this worked quite fast
2008-08-05T15:51:34  <johill> what uses news() already?
2008-08-05T15:54:13  <ThomasWaldmann> lol
2008-08-05T15:54:53  <ThomasWaldmann> johill: there are 2 possibilities: either fix hgcia (not my code) or switch it off
2008-08-05T15:58:00  <johill> I'm fine with switching it off
2008-08-05T15:58:22  <ThomasWaldmann> dennda: as you are already in #cia, did they tell something interesting?
2008-08-05T15:58:43  <dennda> ThomasWaldmann: No response at all
2008-08-05T15:59:24  <dennda> johill: To my knowledge, nothing (except the code is still the same, which I don't believe)
2008-08-05T16:00:04  <ThomasWaldmann> btw, I won't switch it on/off all the time. so if I switch it off now, it will be off for a longer time.
2008-08-05T16:01:21  <johill> dennda: rc doesn't work, does it?
2008-08-05T16:01:34  <johill> dennda: GlobalEditLog uses it
2008-08-05T16:02:27  <dennda> no it doesn't
2008-08-05T16:04:57  <ThomasWaldmann> ehrm, didn't it timeout after 30s?
2008-08-05T16:05:04  <dennda> no
2008-08-05T16:05:11  <dennda> I pushed for at least 10 min
2008-08-05T16:05:39  <ThomasWaldmann> strange, because i have fixed that long ago, but the repo workdir seemed to have some old revision somehow
2008-08-05T16:05:57  <ThomasWaldmann> let's see with next push
2008-08-05T16:18:12  <johill> ok
2008-08-05T16:18:16  * johill adds Revision.timestasmp
2008-08-05T16:18:37  <johill> dennda: can you use that instead of EDIT_LOG_MTIME?
2008-08-05T16:19:11  <dennda> sure
2008-08-05T16:19:17  <dennda> will look into it in a minute
2008-08-05T16:27:26  <johill> somehow EDIT_LOG_ACTION seems to be missing often
2008-08-05T16:29:03  <johill> nah never mind
2008-08-05T16:31:30  <johill> ok, RC works
2008-08-05T16:37:56  <dennda> hah great
2008-08-05T16:38:04  <dennda> I miss CIA-53 telling me you pushed
2008-08-05T16:38:47  <johill> heh
2008-08-05T16:39:08  <dennda> Oh my: http://the-space-station.com/~kaffke/images/lakebennett/00003.jpg
2008-08-05T16:39:31  <dennda> (Australia)
2008-08-05T16:40:00  <johill> heh
2008-08-05T16:40:34  <johill> dennda: you have three test failures now
2008-08-05T16:40:44  <dennda> ok
2008-08-05T16:41:19  <johill> I'll implement revision.size now
2008-08-05T16:41:27  <dennda> great
2008-08-05T16:41:30  <dennda> missed that already
2008-08-05T16:51:21  <ThomasWaldmann> is pushing faster now?
2008-08-05T16:51:37  <ThomasWaldmann> (should have 30s timeout)
2008-08-05T16:51:43  <johill> done
2008-08-05T16:51:49  <johill> ThomasWaldmann: yeah, it seems to time out now
2008-08-05T16:52:51  <johill> dennda: you can use that now
2008-08-05T16:54:19  <dennda> johill: thanks
2008-08-05T16:54:56  <johill> memb has four failing tests now :)
2008-08-05T16:54:59  <johill> you need to implement news()
2008-08-05T16:59:33  <dennda> I know :(
2008-08-05T16:59:42  <dennda> Will need to look at your implementation, though
2008-08-05T17:05:15  <johill> oh fuck
2008-08-05T17:05:34  <johill> I just found a very stupid bug
2008-08-05T17:20:07  <johill> dennda: there, fixed up the form[rev] stuff
2008-08-05T17:26:30  * johill kills all the old storage code
2008-08-05T17:27:19  <dennda> johill is on killing spree!
2008-08-05T17:27:28  <ThomasWaldmann> >:)
2008-08-05T17:27:44  <dennda> (Now it becomes obvious who ever played Quake 3)
2008-08-05T17:28:29  <johill> that stuff was just annoying
2008-08-05T17:28:36  <dennda> true
2008-08-05T17:28:43  <dennda> py.test didn't like it
2008-08-05T17:28:47  <johill> heh
2008-08-05T17:29:36  <johill> have you fixed the caching thing yet?
2008-08-05T17:30:21  <dennda> no
2008-08-05T17:30:37  <johill> is there something like pygrep?
2008-08-05T17:30:42  <dennda> I looked at it once, but that arena stuff was a bit confusing
2008-08-05T17:30:45  <dennda> pygrep?
2008-08-05T17:30:55  <johill> that greps python statements rather than lines
2008-08-05T17:30:55  <dennda> what would you need that for?
2008-08-05T17:31:03  <dennda> ah
2008-08-05T17:31:14  <johill> well I want to find all caching.CacheEntry without scope
2008-08-05T17:31:16  <dennda> dunno, would be cool though
2008-08-05T17:31:17  <johill> but often you have
2008-08-05T17:31:19  <johill> caching.CacheEntry(...,
2008-08-05T17:31:20  <johill> ...
2008-08-05T17:31:23  <johill> scope = 'asdf')
2008-08-05T17:31:34  <dennda> regex-power? :)
2008-08-05T17:31:41  <johill> multiline
2008-08-05T17:31:48  <dennda> so?
2008-08-05T17:36:00  <johill> oi, loads of merge conflicts
2008-08-05T17:38:34  <johill> merged
2008-08-05T17:44:02  <dennda> boy
2008-08-05T17:44:07  <dennda> somebody get CIA-53 working again
2008-08-05T17:44:48  * dennda checks your changesets
2008-08-05T17:46:03  <johill> stupid tests
2008-08-05T18:05:06  *** irc.freenode.net sets mode: +o ChanServ
2008-08-05T18:05:49  <dennda> sigh...
2008-08-05T18:05:58  <johill> ?
2008-08-05T18:08:01  <dennda> You didn't notice there was a huge netsplit, did you? :)
2008-08-05T18:08:26  <johill> I did :)
2008-08-05T18:08:56  <johill> ok... the tests are stupid
2008-08-05T18:09:02  <johill> always assuming a certain underlay is present
2008-08-05T18:10:31  <johill> I guess we should provide them with a memory backend that contains a few pages
2008-08-05T18:10:45  <dennda> anything important I missed?
2008-08-05T18:11:00  <dennda> I was alone with mitsuhiko for some romantic moments :)
2008-08-05T18:12:40  <johill> no
2008-08-05T18:14:53  <johill> you need to go back to Page.py
2008-08-05T18:14:57  <johill> and fix the editlog stuff there
2008-08-05T18:20:28  <dennda> johill: with the latest changeset, does running wikiserver still work for you?
2008-08-05T18:21:23  <dennda> http://paste.pocoo.org/show/81294/
2008-08-05T18:22:47  <dennda> with your fs backend
2008-08-05T18:22:51  <dennda> with memory it works
2008-08-05T18:22:58  <johill> you have old state you need to nuke
2008-08-05T18:23:34  <dennda> yes, just realized that
2008-08-05T18:24:07  <dreimark> it looks like in 1.8 sometimes rename can't remove the cache dir and the edit-log file from a page dir
2008-08-05T18:24:26  <dennda> johill: getting a nosuchrevisionerror now when saving a page
2008-08-05T18:25:27  <dennda> johill: Page.py:558
2008-08-05T18:25:32  <johill> indeed
2008-08-05T18:26:48  <johill> fixing
2008-08-05T18:28:52  <dennda> k
2008-08-05T18:28:54  <dennda> works
2008-08-05T18:29:22  <dennda> johill: Didn't you say RC works now?
2008-08-05T18:29:41  <johill> yes
2008-08-05T18:30:22  <johill> it does here
2008-08-05T18:30:23  <dennda> Doesn't here
2008-08-05T18:30:32  <dennda> and uploading an attachment is also b0rked
2008-08-05T18:31:07  <johill> huh
2008-08-05T18:31:21  <dennda> which is strange, actually
2008-08-05T18:32:03  <johill> works too
2008-08-05T18:32:13  <dennda> hm now it does indeed
2008-08-05T18:32:36  <dennda> anything special I need to do to make RC working?
2008-08-05T18:32:41  <johill> no
2008-08-05T18:32:43  <johill> what happens?
2008-08-05T18:32:48  <dennda> nothing
2008-08-05T18:32:54  <dennda> just shown as regular page
2008-08-05T18:33:02  <johill> have you added <<RecentChanges>>??
2008-08-05T18:33:03  <dennda> (i.e., as it was before)
2008-08-05T18:33:24  <johill> it _is_ just a regular page
2008-08-05T18:33:29  <dennda> ah
2008-08-05T18:33:46  <dennda> now, great
2008-08-05T18:34:27  <johill> there's a weird security test failure
2008-08-05T18:34:50  <johill> I suspect a config issue tho
2008-08-05T18:36:44  <dennda> Damn you DHL, I want my new Screen
2008-08-05T18:39:11  <johill> CacheError: Cannot pickle <type 'thread.lock'> objects
2008-08-05T18:39:38  <johill> wtf?
2008-08-05T18:39:44  <dennda> dunno
2008-08-05T18:40:48  <johill> I suspect some Revision object got added
2008-08-05T18:40:53  <johill> rather than a revno
2008-08-05T18:41:39  <johill> yeah
2008-08-05T18:43:45  <dennda> How can you know that by that error message?
2008-08-05T18:45:06  <johill> where else would a lock object come from?
2008-08-05T18:45:18  <johill> when not from your membackend's revisions or items?
2008-08-05T18:47:38  <dennda> yeah, but still quite sherlock-holmesy
2008-08-05T18:48:46  <johill> :)
2008-08-05T18:49:55  <johill> your Page.getPageList was at fault anyway
2008-08-05T18:49:59  <johill> I fixed it
2008-08-05T18:50:18  <johill> what are you working on?
2008-08-05T18:51:04  <dennda> attachfile
2008-08-05T18:51:45  <johill> Attachment 'friedakv.jpg' does not exist!
2008-08-05T18:51:49  <johill> I just uploaded it?!
2008-08-05T18:52:14  <dennda> hm I had no problems with that up to now and also tried many faulty scenarios
2008-08-05T18:52:23  <johill> get works, but view doesn't
2008-08-05T18:52:24  <dennda> let me check if the latest updates b0rked anything
2008-08-05T18:52:28  <dennda> yes
2008-08-05T18:52:30  <dennda> ah
2008-08-05T18:52:30  <dennda> yes
2008-08-05T18:52:36  <dennda> view isn't rewritten yet :)
2008-08-05T18:52:46  <dennda> and still checks for the file on the disk
2008-08-05T18:53:01  <johill> oh
2008-08-05T18:53:17  <johill> ok the other one then worked again because I uploaded something from my temp dir :)
2008-08-05T18:53:32  <dennda> heh
2008-08-05T18:53:38  <dennda> get and del should work
2008-08-05T18:53:48  <johill> yeah
2008-08-05T18:54:26  <dennda> I need to think how to handle move, though
2008-08-05T18:55:14  <johill> using them with {{ doesnt work either
2008-08-05T18:55:52  <dennda> what does that invoke?
2008-08-05T18:56:06  <johill> MoinMoin/formatter/__init__.py::attachment_inlined
2008-08-05T18:56:15  <dennda> ok
2008-08-05T18:56:19  <johill> AttachFile.absoluteName
2008-08-05T18:56:25  <johill> AttachFile.getFilename
2008-08-05T18:57:19  <johill> and AttachFile.getAttachUrl directly
2008-08-05T18:57:28  <johill> and AttachFile.exists is needed
2008-08-05T19:03:46  <johill> xorAxAx: around?
2008-08-05T19:09:33  <dennda> 'fsize': "%.1f" % float(rev.size) / 1024
2008-08-05T19:09:36  <dennda> What's wrong with that?
2008-08-05T19:10:20  <johill> fsize?
2008-08-05T19:10:32  <dennda> a parameter for building the list
2008-08-05T19:10:40  <johill> no idea
2008-08-05T19:11:04  <dennda> 'fsize': "%.1f" % (float(rev.size) / 1024) <- works
2008-08-05T19:11:17  <dennda> I was under the assumption that / binds stronger than %
2008-08-05T19:12:02  <johill> heh
2008-08-05T19:12:02  <dennda> what the heck
2008-08-05T19:12:04  <dennda> NoSuchRevisionError
2008-08-05T19:12:06  <dennda> again
2008-08-05T19:12:15  <dennda> When uploading an attachment
2008-08-05T19:12:49  <johill> hmm
2008-08-05T19:13:13  <dennda> Want a traceback?
2008-08-05T19:13:32  <johill> sure
2008-08-05T19:14:10  <johill> all fine here tho
2008-08-05T19:14:26  <dennda> http://paste.pocoo.org/show/81297/
2008-08-05T19:14:32  <dennda> actually the problem is in my attachfile
2008-08-05T19:14:33  <dennda> hm
2008-08-05T19:14:38  <dennda> worked before
2008-08-05T19:14:46  <xorAxAx> johill: yes
2008-08-05T19:14:53  <xorAxAx> just arrived from the hermann
2008-08-05T19:16:06  <johill> :)
2008-08-05T19:16:10  <johill> bike?
2008-08-05T19:16:21  <xorAxAx> no, kletterpark :)
2008-08-05T19:17:10  <xorAxAx> "ropes course"
2008-08-05T19:17:14  <xorAxAx> pretty ... challenging
2008-08-05T19:17:56  <johill> ah, yeah, I was there before
2008-08-05T19:23:08  <ThomasWaldmann> johill: thanks for fixing all those pep8
2008-08-05T19:24:23  <johill> xorAxAx: news() now returns just revision objects
2008-08-05T19:24:53  <xorAxAx> johill: hmm, and for renames?
2008-08-05T19:30:31  <johill> yeah exactly
2008-08-05T19:30:34  <johill> I was going to implement that
2008-08-05T19:31:02  <johill> what should that be in there?
2008-08-05T19:31:13  <johill> should we have a (type, data) tuple?
2008-08-05T19:32:35  <ThomasWaldmann> johill: isn't there something missing?:
2008-08-05T19:32:37  <ThomasWaldmann> - return self._item.get_revision(-1)[ACL]
2008-08-05T19:32:37  <ThomasWaldmann> + text = self._item.get_revision(-1)[ACL]
2008-08-05T19:32:38  <ThomasWaldmann> + AccessControlList(self.request.cfg, text)
2008-08-05T19:33:46  <xorAxAx> johill: so you have to do isinstance everytime?
2008-08-05T19:35:29  <ThomasWaldmann> +        for key in self._user.keys():
2008-08-05T19:35:38  <ThomasWaldmann> and can't we just have __contains__?
2008-08-05T19:37:25  <xorAxAx> that looks indeed incomplete procotol wise :)
2008-08-05T19:40:54  <johill> ThomasWaldmann: hm missing?
2008-08-05T19:41:16  <ThomasWaldmann> just have a 2nd look :)
2008-08-05T19:41:23  <johill> what does this have to do with __contains__?
2008-08-05T19:41:34  <ThomasWaldmann> (at the acl stuff)
2008-08-05T19:41:39  <xorAxAx> two different hunks, johill
2008-08-05T19:41:41  <johill> the acl stuff was just wrong
2008-08-05T19:41:46  <johill> I made a mistake the first time
2008-08-05T19:41:48  <ThomasWaldmann> and I meant __iter__ or something
2008-08-05T19:41:52  <johill> and then fixed it up, you have to look at both patches
2008-08-05T19:42:20  <johill> oh. missing return, heh
2008-08-05T19:42:21  <ThomasWaldmann> that was from 2nd patch
2008-08-05T19:42:48  <ThomasWaldmann> and I also suggest to compare it with 1.8 main branch
2008-08-05T19:43:07  <ThomasWaldmann> the acl code had quite some brokenness that was fixed not too long ago
2008-08-05T19:43:19  <johill> I merged the main branch
2008-08-05T19:43:26  <johill> I should have gotten conflicts if there was anything
2008-08-05T19:43:46  <ThomasWaldmann> (the acl hierarchic code was very broken)
2008-08-05T19:43:52  <johill> and I would appreciate if you just said what is missing instead of forcing me to look at the code a dozen times
2008-08-05T19:44:00  <johill> I'm not a in a "needs to learn python" state any more, thank you
2008-08-05T19:44:21  <ThomasWaldmann> maybe just be a bit more careful, esp. if you change acl code
2008-08-05T19:44:47  <johill> the tests all pass now
2008-08-05T19:45:05  <johill> no actually, they don't
2008-08-05T19:45:09  <johill> look
2008-08-05T19:45:14  <johill> the damn code was just totally wrong before
2008-08-05T19:45:18  <johill> give me a break
2008-08-05T19:45:39  <johill> if everybody was more careful I wouldn't have to fix your test failures either
2008-08-05T19:47:57  <johill> the "for key in user.keys()" also has nothing to do with __contains__
2008-08-05T19:47:59  <xorAxAx> OMFG, XSS in hg's rss feed
2008-08-05T19:48:11  <xorAxAx> and XSS in hgweb generally
2008-08-05T19:48:21  <xorAxAx> hmm, or is it a feature
2008-08-05T19:48:24  <johill> for key in item: works fine, but will normally not work while you modify the dict at the same itme
2008-08-05T19:48:36  <xorAxAx> ah, a feature
2008-08-05T19:48:57  <johill> xorAxAx: so how would you like item renames to show up in news?
2008-08-05T19:49:10  <johill> xorAxAx: a new class and isinstance?
2008-08-05T19:49:31  <xorAxAx> johill: well, if you dislike the idea of the meta-item, then you would probably just yield objects and people need isinstance/getattr
2008-08-05T19:49:57  <xorAxAx> and probably you can even think about a class/interface that you mixin into rev/rename instances
2008-08-05T19:50:03  <xorAxAx> to get some common methods
2008-08-05T19:50:08  <xorAxAx> (no idea if that makes sense9
2008-08-05T19:50:41  <johill> I don't see them having any common stuff except maybe timestamp
2008-08-05T19:51:42  <johill> and an item pointer
2008-08-05T19:53:14  <johill> but I suppose renames would have something like Rename.from_name, Rename.to_name
2008-08-05T19:53:32  <xorAxAx> yeah, probably
2008-08-05T19:53:48  <johill> and Rename.item
2008-08-05T19:53:58  <johill> which is the actual item, in case you need it
2008-08-05T19:54:04  <johill> (might have been renamed again, after all)
2008-08-05T19:54:15  * dennda feels sick
2008-08-05T19:56:31  <xorAxAx> johill: doesnt make sense to provide that attrib :
2008-08-05T19:56:46  <johill> why not? we can easily have the data to fill it
2008-08-05T19:56:48  <xorAxAx> except if you include some transitive history traversal
2008-08-05T19:57:07  <johill> what do you mean?
2008-08-05T19:57:30  <xorAxAx> that the item property traverses the news() and returns the current name of the item
2008-08-05T19:57:33  <xorAxAx> hmm
2008-08-05T19:57:39  <xorAxAx> or better, the current item
2008-08-05T19:57:53  <johill> well it'd be easy to get the current item
2008-08-05T19:58:00  <johill> if we encode it into the api
2008-08-05T19:58:23  <xorAxAx> yes, the idea is just that item shouldnt just return backend.getitem(to_name) :)
2008-08-05T19:58:33  <johill> sure
2008-08-05T19:58:42  <johill> but it's easy
2008-08-05T19:58:53  <johill> the backend can store the ID or whatever
2008-08-05T19:58:56  <xorAxAx> yeah, but O(n)
2008-08-05T19:59:01  <xorAxAx> hmm
2008-08-05T19:59:18  <xorAxAx> the base class could have a linear implementation
2008-08-05T19:59:22  <xorAxAx> for simple bacends
2008-08-05T19:59:35  <johill> I don't see how
2008-08-05T20:00:05  <xorAxAx> just build the transitive closure of all rename ops in news(renameobj.timestamp)
2008-08-05T20:00:16  <xorAxAx> starting from the to_name
2008-08-05T20:00:21  <johill> that's racy
2008-08-05T20:01:16  <xorAxAx> yeah
2008-08-05T20:01:24  <xorAxAx> if somebody just renames ...
2008-08-05T20:01:37  <xorAxAx> but dont you lock your cdb file anyway? :)
2008-08-05T20:01:55  <xorAxAx> just joking :)
2008-08-05T20:02:06  <xorAxAx> hmm, but not sure whether it makes sense to ask every backend to implement it
2008-08-05T20:02:09  <johill> not for reading. if I want to implement this for the fs backend, I'd just store the ID
2008-08-05T20:02:17  <johill> no big deal
2008-08-05T20:02:22  <xorAxAx> probably its better to provide a racy version for simple backends
2008-08-05T20:02:28  <johill> I disagree
2008-08-05T20:02:30  <xorAxAx> those that dislike indirections of internal ids :)
2008-08-05T20:02:39  <xorAxAx> ok, i dont care enough
2008-08-05T20:02:41  <johill> I don't want any racy default code
2008-08-05T20:02:44  <xorAxAx> moin is enterprisey anyway :)
2008-08-05T20:02:50  <xorAxAx> hah, there is enough of it :)
2008-08-05T20:02:59  <johill> not in storage :)
2008-08-05T20:03:05  <xorAxAx> yeah
2008-08-05T20:04:07  <johill> right now though, I don't even see how to save this data
2008-08-05T20:04:16  <johill> I'll figure it out :)
2008-08-05T20:04:58  <johill> something like <oldname, newname, itemid, recordlength> will probably work
2008-08-05T20:05:29  <johill> with oldname being empty for new items
2008-08-05T20:12:52  <johill> hmm
2008-08-05T20:12:53  <johill> otoh
2008-08-05T20:12:59  <johill> why does the item log have to be interleaved into news anyway?
2008-08-05T20:13:05  <johill> that's trivial to do for code that requires that
2008-08-05T20:13:21  <xorAxAx> interleaved?
2008-08-05T20:13:40  <xorAxAx> the task is to display RC without any O(#items) component in the complexity
2008-08-05T20:13:48  <johill> sure
2008-08-05T20:14:00  <xorAxAx> so what do you mean?
2008-08-05T20:14:10  <johill> it's trivial to interleave two iterators based on timestamps
2008-08-05T20:14:20  <xorAxAx> yeah, but probably in the backend
2008-08-05T20:14:37  <xorAxAx> because i dont see the case in which you only need one of them :)
2008-08-05T20:14:46  <johill> heh
2008-08-05T20:15:27  <johill> well not really actually
2008-08-05T20:15:34  <johill> RC doesn't need renames
2008-08-05T20:15:40  <xorAxAx> sure it does
2008-08-05T20:15:46  <johill> no
2008-08-05T20:15:48  <xorAxAx> you want to show people that a page was renamed
2008-08-05T20:15:52  <johill> the frontend code creates new revisions
2008-08-05T20:15:55  <xorAxAx> not that it was edited
2008-08-05T20:16:11  <johill> sure
2008-08-05T20:16:23  <xorAxAx> so how do you find out that a page was renamed?
2008-08-05T20:16:49  <johill> well RC doesn't need to find out because the rename code adds a revision with "this was renamed"
2008-08-05T20:17:19  <xorAxAx> ?
2008-08-05T20:17:27  <xorAxAx> that doesnt help you to show it in the RC
2008-08-05T20:17:36  <xorAxAx> or do you want to parse the first line of the rev? :)
2008-08-05T20:17:41  <johill> umm
2008-08-05T20:17:48  <johill> the data for that is already in the metadata for the revision
2008-08-05T20:17:50  <johill> look
2008-08-05T20:17:56  <xorAxAx> hmm
2008-08-05T20:17:56  <johill> check out the branch and just try :)
2008-08-05T20:18:07  <xorAxAx> so the name is also stored in the comment?
2008-08-05T20:18:11  <xorAxAx> sounds dirty to me IMHO
2008-08-05T20:18:22  <xorAxAx> to make this the preferred way
2008-08-05T20:18:24  <johill> no
2008-08-05T20:18:33  <johill> there's edit log comment and edit log extra
2008-08-05T20:18:33  <xorAxAx> there is no name? even worse! :)
2008-08-05T20:18:38  <johill> you suck
2008-08-05T20:18:40  <xorAxAx> :-)
2008-08-05T20:18:44  <johill> go understand anything before making comments
2008-08-05T20:18:56  <johill> it'd even help if you'd understand how it works now :)
2008-08-05T20:19:00  <xorAxAx> i think my confusing point is still valid
2008-08-05T20:19:16  <johill> confusing point?
2008-08-05T20:19:21  <dennda> lol, be nice guys :)
2008-08-05T20:19:34  <xorAxAx> we might want to show the old page name. if we store it in the rev metadata, we duplicate it from the meta data that the bacend records anyway
2008-08-05T20:19:52  <xorAxAx> johill: even better if it wasnt confusing
2008-08-05T20:20:05  <johill> the old page name is shown just fine
2008-08-05T20:20:17  <xorAxAx> yes, i understood that :)
2008-08-05T20:20:18  <johill> it's part of the metadata the frontend code puts into the new revision as "edit log"
2008-08-05T20:20:35  <xorAxAx> does that make sense?
2008-08-05T20:20:37  <johill> the backend records, currently, no meta data about renames
2008-08-05T20:20:47  <xorAxAx> that there is a name in the rev metadata
2008-08-05T20:20:57  <johill> well it's how the frontend code works, no, it probably doesn't make sense to create a new revision in the first place
2008-08-05T20:21:02  <xorAxAx> (i think that is what it boils down to)
2008-08-05T20:21:39  <xorAxAx> well, the frontend code could just operate on the real backend metadata instead of taking it from the denormalized place
2008-08-05T20:21:47  <xorAxAx> without any performance hit
2008-08-05T20:22:45  <johill> if we add such name logging metadata, then we should remove from renaming the new revision part
2008-08-05T20:23:16  <johill> that's a good thing, its racy anyway
2008-08-05T20:23:54  <xorAxAx> hmm, whats racy?
2008-08-05T20:24:31  <xorAxAx> the metadata of the rev just says "the item i am belonging to has this name now"
2008-08-05T20:24:48  <xorAxAx> but the item also has a current name and the backed is logging changes of the name
2008-08-05T20:24:53  <johill> the racy thing is that if somebody tries to edit the thing at the same time you rename it
2008-08-05T20:25:22  <johill> you may have renamed it, but the edit is overwritten by the new rename revision
2008-08-05T20:25:40  <johill> i.e. the edit is lost
2008-08-05T20:25:50  <xorAxAx> huh
2008-08-05T20:26:17  <johill> can we agree that the "## page was renamed from XX" is useless?
2008-08-05T20:26:21  <xorAxAx> yes
2008-08-05T20:26:22  <xorAxAx> you say that there is no atomic racefree append?
2008-08-05T20:26:37  <xorAxAx> i mean, there might be useful actions that want to add a line in the wikipage
2008-08-05T20:26:56  <johill> there's no atomic rename+addrevision
2008-08-05T20:27:18  <xorAxAx> so editing wiki pages is racy if somebody overrides the highlevel editlock (which just means ENTER)?
2008-08-05T20:27:37  <xorAxAx> (hmm, it currently is too :)
2008-08-05T20:27:43  <johill> no
2008-08-05T20:28:15  <johill> in fact, I just fixed it
2008-08-05T20:28:28  <johill> (you'll get an exception when a race happened, dennda has to fix that)
2008-08-05T20:31:46  <johill> xorAxAx: you just cannot do both things at the same time atomically
2008-08-05T20:32:52  <xorAxAx> well, you have to retry
2008-08-05T20:33:12  <johill> but we don't actually need to at all if we don't want '## renamed from foo' on the page
2008-08-05T20:33:29  <xorAxAx> think about a comment action
2008-08-05T20:33:34  <xorAxAx> or just editing pages
2008-08-05T20:33:39  <johill> ???
2008-08-05T20:33:43  <johill> you still haven't understood it
2008-08-05T20:33:54  <johill> why would a comment action try to rename the page?
2008-08-05T20:33:56  <xorAxAx> ok, the edit action has the merging code in front
2008-08-05T20:34:07  <xorAxAx> johill: you have an edit-edit race as well
2008-08-05T20:34:21  <xorAxAx> if the code is broken
2008-08-05T20:34:28  <johill> well edit action will use the revno from the form to avoid it
2008-08-05T20:34:34  <johill> and a comment action probably should too
2008-08-05T20:34:52  <xorAxAx> why? it just needs to append to the end of the page
2008-08-05T20:34:59  <xorAxAx> but it might have to retry internally
2008-08-05T20:35:05  <johill> in that case it can just retry, sure
2008-08-05T20:35:07  <xorAxAx> but that works already, i know
2008-08-05T20:35:21  <johill> so how is all that related to our rename discussion?
2008-08-05T20:35:22  <xorAxAx> (just that nobody has implemented that yet, one would have done so in the rename code, though)
2008-08-05T20:35:31  <xorAxAx> not at all
2008-08-05T20:36:10  <johill> so if we want backend-level rename logs, we should imho remove the new revision thing anyway
2008-08-05T20:36:53  <xorAxAx> yes, as we agreed on a few hours ago :)
2008-08-05T20:37:55  <johill> :)
2008-08-05T20:42:31  <xorAxAx> another central idea of it was to get mails sent out
2008-08-05T20:42:39  <xorAxAx> but that works differently nowadays anyway
2008-08-05T20:42:52  <johill> heh
2008-08-05T20:46:03  <johill> hmmm
2008-08-05T20:46:13  <johill> if we store this just in the item log, then the info action won't show renames
2008-08-05T20:46:26  <xorAxAx> yeah, probably fine
2008-08-05T20:46:41  <xorAxAx> data model wise, annoying anyway
2008-08-05T20:47:25  <xorAxAx> (but in which cases?)
2008-08-05T20:47:32  <xorAxAx> hmm, we should ask the users
2008-08-05T20:47:44  <johill> heh
2008-08-05T20:47:46  * xorAxAx wouldnt expect the info actio to show it
2008-08-05T20:48:00  <johill> it does now, I couldn't care less I guess
2008-08-05T21:34:47  <johill> ah crap
2008-08-05T21:35:15  <johill> we need to store "who did it" and stuff like that with rena,es
2008-08-05T21:35:37  * johill gives up for the day
2008-08-05T21:35:52  * dennda carefully removes EDIT_LOG_MTIME
2008-08-05T21:38:07  <ThomasWaldmann> hm?
2008-08-05T21:38:52  <johill> hm?
2008-08-05T21:39:00  <johill> ThomasWaldmann: what?
2008-08-05T21:39:20  <ThomasWaldmann> dennda: why do you remove storage of MTIME?
2008-08-05T21:39:28  <johill> he doesn't
2008-08-05T21:39:36  <johill> you haven't followed the discussion
2008-08-05T21:39:49  <ThomasWaldmann> not all at least :)
2008-08-05T21:40:03  <dennda> Not the important part :)
2008-08-05T21:40:07  <johill> the mtime is now revison.timestamp rather than revision[editlogmtime]
2008-08-05T21:40:20  <johill> because of reasons that you'll have to read yourself ;)
2008-08-05T21:43:05  <dennda> Which is much better by the way
2008-08-05T21:43:17  <dennda> Keeps me from doing try: ... except KeyError...
2008-08-05T21:44:27  <TheSheep> .get(..., some default)
2008-08-05T21:44:48  <dennda> TheSheep: oh, right ;)
2008-08-05T21:45:05  * ThomasWaldmann .oO(many pages from the 70s :)
2008-08-05T21:45:08  <dennda> I know that but havn't used it myself.
2008-08-05T21:45:10  <TheSheep> dennda: so now it'll be AttributeError? ;)
2008-08-05T21:45:28  <dennda> TheSheep: No, it's always there :)
2008-08-05T21:48:16  <xorAxAx> johill: another reason to use a meta item :)
2008-08-05T21:48:41  <johill> xorAxAx: yeah, I'm starting to swing that way
2008-08-05T21:49:32  <johill> xorAxAx: except it's not really atomic then, is it?
2008-08-05T21:50:08  <xorAxAx> johill: hmm
2008-08-05T21:50:53  <johill> i.e. you'd either have to have the backend do it (which means the new API needs to be added anyway) or do some weird high-level locking or so
2008-08-05T21:51:10  <xorAxAx> yes, the backend has to write these entries
2008-08-05T21:52:01  <johill> that means, however, that the rename API has to get a metadata dict parameter
2008-08-05T21:53:42  <xorAxAx> hmm, indeed
2008-08-05T21:53:46  <xorAxAx> layering violation
2008-08-05T21:54:00  * TheSheep notes that just over a week to a deadline is not really the right time to change the spec
2008-08-05T21:54:06  <xorAxAx> which spec? :)
2008-08-05T21:54:35  <TheSheep> xorAxAx: the one in johill's head
2008-08-05T21:54:47  <johill> not necessarily a layering violation, I guess, but... bbl
2008-08-05T21:56:45  <xorAxAx> johill: indeed
2008-08-05T22:09:43  <johill> xorAxAx: however, if we pass arbitrary metadata to rename, then it doesn't matter how it's stored
2008-08-05T22:10:28  <xorAxAx> yes
2008-08-05T22:29:53  <johill> xorAxAx: also, when a new item is created by way of publishing metadata for it, is that logged? or just renames?
2008-08-05T22:30:19  <johill> for the wiki, just renames are enough, I guess, since you cannot rename an item without revisions (and if you actually did somehow nobody cares)
2008-08-05T22:31:44  <xorAxAx> well, RC knows that and filters those items
2008-08-05T22:32:37  <johill> yeah but do we need them in the first place?
2008-08-05T22:33:51  <xorAxAx> yes, as long as you can create items without revs
2008-08-05T22:34:01  <xorAxAx> imagine that you want to list all users by their sign on date
2008-08-05T22:34:07  <xorAxAx> like in sysadmin
2008-08-05T22:34:39  <johill> hm ok, although for  that we probably use some actual metadata now
2008-08-05T22:35:20  <johill> I suppose for that I should add Item.timestamp
2008-08-05T22:44:29  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4035:d40461b9ee6c 1.8-wsgi-fkrupicka/contrib/flup-server/ (24 files in 4 dirs): Imported a current flup with single-thread server patch applied
2008-08-05T22:44:30  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4036:fa648504e1cf 1.8-wsgi-fkrupicka/MoinMoin/web/profile.py: Add profile/callgraph functionality from MoinMoin.server.server_standalone as WSGI middlewares
2008-08-05T22:44:31  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4037:dc253f097867 1.8-wsgi-fkrupicka/MoinMoin/web/serving.py: Split application preparation (middlewares) from running a server
2008-08-05T22:44:32  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4038:65af466e6df0 1.8-wsgi-fkrupicka/ (3 files in 2 dirs): Combined deployment script for CGI/FCGI
2008-08-05T22:44:35  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4039:1e21b63603f7 1.8-wsgi-fkrupicka/MoinMoin/ (failure.py support/cgitb.py): Remove deprecated traceback render code
2008-08-05T22:44:38  <CIA-53> Florian Krupicka <florian.krupicka@googlemail.com> default * 4040:63c1baacceea 1.8-wsgi-fkrupicka/MoinMoin/ (3 files in 3 dirs): Remove deprecated FCGI code
2008-08-05T22:50:28  <CIA-53> Johannes Berg <johannes AT sipsolutions DOT net> default * 4562:ebf7a81e68fb 1.8-storage-cdenter/MoinMoin/macro/RecentChanges.py: recentchanges: global edit log already uses int revnos
2008-08-05T22:55:29  <johill> ugh
2008-08-05T22:55:34  <johill> I think I need to rewrite this as iterators
2008-08-05T22:55:42  <johill> generators
2008-08-05T23:07:15  <zenhase> :)
2008-08-05T23:09:27  <ThomasWaldmann> zenhase: /usr/share/moin/htdocs only works when moin is installed with setup.py
2008-08-05T23:10:34  <ThomasWaldmann> (for shared is True)
2008-08-05T23:10:54  <johill> dennda: deleting pages is totally messed up
2008-08-05T23:13:11  <ThomasWaldmann> zenhase: cgi is done by same script as fcgi??
2008-08-05T23:15:05  <ThomasWaldmann> 44        On most platforms, fcgi will fallback to regular CGI behavior if run in a
2008-08-05T23:15:06  <ThomasWaldmann> 45        non-FastCGI context.
2008-08-05T23:15:12  <ThomasWaldmann> nice
2008-08-05T23:15:49  <zenhase> yep
2008-08-05T23:16:12  <ThomasWaldmann> the logging configuration MUST be done before any loggers are used
2008-08-05T23:16:20  <zenhase> hmm ok
2008-08-05T23:17:16  <zenhase> pythons logging-framework doesn't work with handlers added after first log-attempts?
2008-08-05T23:17:19  <ThomasWaldmann> ah, it is still there in moin.cgi
2008-08-05T23:17:26  <zenhase> yes
2008-08-05T23:17:29  <dennda> johill: Please elaborate
2008-08-05T23:17:42  <johill> dennda: well it doesn't show "page doesn't exist" for one
2008-08-05T23:17:46  <ThomasWaldmann> zenhase: it does weird magic behind the scenes
2008-08-05T23:17:47  <zenhase> though i am thinking about putting some autodetection-code somewhere
2008-08-05T23:17:52  <johill> dennda: and ?action=raw shows the last data
2008-08-05T23:18:12  <zenhase> that tries to find logging and wikiconfig in some expected places
2008-08-05T23:18:28  <dennda> ok
2008-08-05T23:19:09  <zenhase> but i'll go home now and go to sleep i think and then continue tomorrow
2008-08-05T23:19:17  <ThomasWaldmann> zenhase: you aren't using failure any more, no callers left?
2008-08-05T23:19:30  <zenhase> grepping didn't find any
2008-08-05T23:19:34  <ThomasWaldmann> ok
2008-08-05T23:19:52  <ThomasWaldmann> and instead cgitb, we have some nice werkzeug stuff now?
2008-08-05T23:19:52  <zenhase> and cgitb/failure is kind of replaceable with werkzeugs DebugApplication
2008-08-05T23:19:57  <zenhase> yes
2008-08-05T23:20:30  <zenhase> nice rendering and has support to evaluate code in the different traceback-frames for testing
2008-08-05T23:20:33  <ThomasWaldmann> do you log all exceptions?
2008-08-05T23:20:49  <ThomasWaldmann> using logging.exception
2008-08-05T23:21:00  <zenhase> hmm
2008-08-05T23:21:54  <johill> PawelPacana: ping
2008-08-05T23:22:47  <zenhase> ThomasWaldmann: werkzeugs DebugApplication logs exceptions to wsgi.error
2008-08-05T23:23:23  <zenhase> oh no, i am wrong
2008-08-05T23:23:38  <zenhase> that is only when response data was already written to the client
2008-08-05T23:24:12  <zenhase> i will have a look and incorporate it at an appropiate place
2008-08-05T23:25:47  <ThomasWaldmann> zenhase: that except ImportError is evil :)
2008-08-05T23:25:55  <ThomasWaldmann> what do you plan there?
2008-08-05T23:26:50  <zenhase> hmm, i wondered if i can somehow give a appropiate error-message to the user
2008-08-05T23:27:11  <zenhase> but for fcgi this won't be possible i suppose :o
2008-08-05T23:27:23  <dreimark> ++
2008-08-05T23:28:03  <zenhase> since i need to import fcgi to render an error to the client about not being able to import fcgi ;)
2008-08-05T23:28:24  <TheSheep> zenhase: didn't werkzeug have an exception-catching middleware, similar to cgitb?
2008-08-05T23:28:47  <zenhase> TheSheep: yes it has
2008-08-05T23:29:02  <zenhase> TheSheep: DebuggedApplication, just mentioned above
2008-08-05T23:29:16  <zenhase> way beyond cgitb even
2008-08-05T23:29:41  <TheSheep> ah, sorry, I didn't pay attention
2008-08-05T23:29:44  <zenhase> more features (evaluation in traceback frames, post-to-paste.pocoo.org, etc.)
2008-08-05T23:29:50  <TheSheep> just a 25-line screen :)
2008-08-05T23:29:57  <zenhase> hehe :)
2008-08-05T23:30:03  * zenhase donates some lines to TheSheep 
2008-08-05T23:30:45  <zenhase> now, biking home ->
2008-08-05T23:31:43  <PawelPacana> johill: pong
2008-08-05T23:31:59  <johill> PawelPacana: is there any way to run your backend without needing some vcs version of hg?
2008-08-05T23:32:10  <johill> PawelPacana: can you add appropriate skips if the version isn't sufficiently new?
2008-08-05T23:32:31  <PawelPacana> i have this on my todo (skips)
2008-08-05T23:32:59  <PawelPacana> no way to run on non-devel (hg API changed)
2008-08-05T23:33:43  <TheSheep> johill: by the time it's in moin, the dev version of hg will be in debian stable ;)
2008-08-05T23:34:10  <johill> TheSheep: do I care? point is that I cannot help you guys at all now
2008-08-05T23:34:21  <TheSheep> johill: why not?
2008-08-05T23:34:32  <johill> because I'd have to learn more about hg than I want to know
2008-08-05T23:35:57  <TheSheep> for example how t run hg clone and setup.py with --prefix option, right
2008-08-05T23:36:23  <TheSheep> johill: want me to tar up the module and put it somewhere for download?
2008-08-05T23:36:27  <johill> and how then how to get it into the right python path in moin, etc.
2008-08-05T23:36:33  <johill> not really
2008-08-05T23:36:38  <TheSheep> johill: you just have to put it in the same dir where MoinMoin is
2008-08-05T23:36:47  * dreimark would like to see a description for that too
2008-08-05T23:36:57  <TheSheep> johill: the path is set in wikiconfig.py anyways
2008-08-05T23:37:49  <PawelPacana> dreimark: i'll post it in a minute
2008-08-05T23:37:55  <johill> 19G   19G  123M 100% /home/johannes
2008-08-05T23:38:21  <PawelPacana> 21M  _workspace/mercurial-devel/
2008-08-05T23:38:27  * ThomasWaldmann also filled his home today
2008-08-05T23:39:04  <TheSheep> sheep@ghostwheel:~/dev/moin/1.8-mercurialbackend-ppacana$ du -s mercurial/ hgext/
2008-08-05T23:39:07  <TheSheep> 2068    mercurial/
2008-08-05T23:39:09  <TheSheep> 1208    hgext/
2008-08-05T23:39:31  <TheSheep> that's whooping 3MB
2008-08-05T23:39:48  <PawelPacana> ?
2008-08-05T23:39:49  <johill> that's pretty good, moin is like 80
2008-08-05T23:40:06  * ThomasWaldmann offers 2 floppies :)
2008-08-05T23:40:16  <johill> anyway, too much hassle
2008-08-05T23:41:01  <johill> PawelPacana: make sure when you support the timestmap you also support setting it to something historic for conversions
2008-08-05T23:41:55  <TheSheep> anyways, http://sheep.art.pl/misc/hg-dev.tar.gz <-- you just have to unpack it in moin's main dir, at tarball's root
2008-08-05T23:42:17  <xorAxAx> well, there is an obvious difference between debugapplication and cgitb
2008-08-05T23:42:33  <xorAxAx> cgitb is currently used in all installations while debugapplication may not be used at all by non-devs
2008-08-05T23:42:38  <TheSheep> xorAxAx: that's nice
2008-08-05T23:42:47  <xorAxAx> (security-wise)
2008-08-05T23:42:52  <xorAxAx> so bugreports will likely have less information
2008-08-05T23:43:12  <xorAxAx> (if it is replaced)
2008-08-05T23:43:27  <xorAxAx> but they can be combined well i think
2008-08-05T23:43:38  <TheSheep> xorAxAx: it can't be enabled by default?
2008-08-05T23:43:43  <xorAxAx> TheSheep: debugapp? no
2008-08-05T23:43:53  <xorAxAx> it allows arbitrary code execution AFAIK
2008-08-05T23:45:06  <dreimark> TheSheep: how to configure ?
2008-08-05T23:45:27  <dreimark> wikiconfig_local ?
2008-08-05T23:47:56  <TheSheep> dreimark: I don't think the upper layers are ready to actually run the wiki on it yet
2008-08-05T23:48:18  <TheSheep> dreimark: but it changes hourly, so I may be wrong
2008-08-05T23:48:53  <dreimark> currently I can't try because I don't know how to ommit "error in config: no user storage configured"
2008-08-05T23:49:08  <johill> user_backend = something
2008-08-05T23:49:32  <dreimark> I know but what can I choose for something
2008-08-05T23:49:37  <ThomasWaldmann> zenhase: server/server_cgi server_modpython (and server_wsgi if you have all functionality in your code also) can get removed now, I guess
2008-08-05T23:49:43  <CIA-53> Pawel Pacana <pawel.pacana@gmail.com> default * 4599:cfe5ac61a37c 1.8-mercurialbackend-ppacana/MoinMoin/storage/backends/hg.py: (hg) Docstring update: mercurial-devel installation instructions.
2008-08-05T23:50:06  <johill> ImportError: /home/johannes/Projects/moin/1.8-mercurialbackend-ppacana/mercurial/bdiff.so: ELF file data encoding not big-endian
2008-08-05T23:50:25  <dennda> dreimark: from MoinMoin.storage.backends import memory, fs; user_backend = memory.MemoryBackend()
2008-08-05T23:50:41  <TheSheep> johill: right, then it's not Intel :)
2008-08-05T23:51:02  <ThomasWaldmann> zenhase: request/request_cgi request_modpython (and maybe request_wsgi) also?
2008-08-05T23:51:07  <TheSheep> johill: what arch, maybe I'll compile it for you
2008-08-05T23:51:10  <johill> obviously it's not python ;)
2008-08-05T23:51:14  <johill> don't bother
2008-08-05T23:51:24  <PawelPacana> dreimark: pull changes, instructions in docstring
2008-08-05T23:51:53  * TheSheep stops bothering
2008-08-05T23:55:33  <dreimark> doing
2008-08-05T23:57:33  <xorAxAx> johill: moin has the bdiff module in its pythonic version
2008-08-05T23:57:40  <xorAxAx> (just as a sidemark :))