MoinMoin: MoinMoinChat/Logs/moin-dev/2007-05-10

2007-05-10T03:24:26  <grzywacz> It's scary how much better I feel working at night. ;S
2007-05-10T03:37:08  <starshine> maybe not
2007-05-10T03:37:15  <starshine> you allergic to something that's up days?
2007-05-10T03:37:25  <starshine> perfume on a co worker?  daytime plants?
2007-05-10T03:37:32  <starshine> nasty AC setup?
2007-05-10T03:38:10  * starshine recodes the sinuses for better exception handling :P
2007-05-10T03:38:47  <grzywacz> Light? Noise of from outside? ;)
2007-05-10T03:39:01  <grzywacz> s/of//
2007-05-10T03:39:11  <starshine> sure
2007-05-10T03:39:20  <grzywacz> No idea.
2007-05-10T03:39:21  <starshine> many excellent programmers are nbight owl types :D
2007-05-10T03:39:28  <grzywacz> hah
2007-05-10T03:39:45  <grzywacz> I'm not excellent, but I enjoy working in my dark room. ;-)
2007-05-10T04:06:17  <ThomasWaldmann> moin grzywacz :)
2007-05-10T04:06:33  <grzywacz> moin ThomasWaldmann :-)
2007-05-10T04:10:00  <grzywacz> ThomasWaldmann, have you just woken up, or do you stay awake this late as well? ;)
2007-05-10T04:16:21  <ThomasWaldmann> i have just woken up in this case :)
2007-05-10T04:16:56  <grzywacz> At 4am? Strange. ;)
2007-05-10T04:22:26  <ThomasWaldmann> sometimes I also work all night (when in coding mood :)
2007-05-10T04:22:51  <grzywacz> 8)
2007-05-10T04:28:38  <grzywacz> ThomasWaldmann, do you happen to have some experience with prolog?
2007-05-10T04:43:56  <grzywacz> I'm going to take a short nap, 'cos I have to wake up in about two hours to go to uni. :D
2007-05-10T08:07:55  <ThomasWaldmann> grzywacz: yes, but it was long ago
2007-05-10T09:14:39  <dreimark> moin
2007-05-10T10:17:30  <lanius> xorAxAx: ping
2007-05-10T10:17:37  <xorAxAx> pong :)
2007-05-10T10:17:41  <lanius> moin :-)
2007-05-10T10:17:49  <xorAxAx> twiki :-)
2007-05-10T10:18:08  <lanius> mediawiki...
2007-05-10T10:18:43  <lanius> "How can a user of a Revision get the data file for writing or for reading?" this is regarding the missing open method now?
2007-05-10T10:19:44  <xorAxAx> no
2007-05-10T10:19:53  <xorAxAx> because there is only one data attribute
2007-05-10T10:20:05  <lanius> which has all required methods
2007-05-10T10:20:49  <xorAxAx> but you cannot offer the same object because it would have to do security checking which is a layering violation
2007-05-10T10:21:32  <lanius> but it could be a ReadableData when you only have read rights where the write methods throws a ACLError
2007-05-10T10:22:03  <xorAxAx> but write is truncating to 0 bytes ...
2007-05-10T10:22:05  <xorAxAx> as we defined
2007-05-10T10:22:13  <xorAxAx> so you cant read from it
2007-05-10T10:22:28  <lanius> but it is only written on close
2007-05-10T10:22:42  <lanius> or on save
2007-05-10T10:22:57  <lanius> and you read the data that is already saved
2007-05-10T10:23:36  <xorAxAx> no, we said that this will make the implementation more expensive and difficult
2007-05-10T10:24:03  <xorAxAx> IMHO it should only be possible to read or write to keep it simple
2007-05-10T10:24:24  <lanius> i don't see the difference, when you have separate objects you do the same and the aim is to have an interface as simple as possible
2007-05-10T10:24:50  <xorAxAx> the difference is that you have to spill the blob to disk if you want to offer reading and writing
2007-05-10T10:24:59  <xorAxAx> i think i have already said that
2007-05-10T10:25:35  <lanius> i don't see the difference if you use two objects
2007-05-10T10:25:55  <xorAxAx> the read object doesnt have to spill the blob to disk
2007-05-10T10:26:19  <lanius> the one object only need to spill on write operations, not on read
2007-05-10T10:26:24  <lanius> as well
2007-05-10T10:27:03  <xorAxAx> the write object would not need to spill at all
2007-05-10T10:27:18  <xorAxAx> because its implicitly truncating
2007-05-10T10:27:47  <lanius> you just said that the read object doesn't have to spill, so i assumed that the write object hast to
2007-05-10T10:27:52  <lanius> our 4gb example...
2007-05-10T10:28:12  <xorAxAx> spill == db to drive
2007-05-10T10:28:25  <lanius> ?
2007-05-10T10:28:31  <xorAxAx> with your object, you would need to read the old file from db just to implement write + read
2007-05-10T10:28:40  <lanius> no
2007-05-10T10:28:45  <xorAxAx> because the read could read from places that havent been written
2007-05-10T10:28:49  <xorAxAx> why?
2007-05-10T10:29:09  <xorAxAx> then describe how you would implement this sequence:
2007-05-10T10:29:18  <lanius> the read object reads the data from the database when it is accessed and holds it in mem or disk depending on the size
2007-05-10T10:29:29  <xorAxAx> seek(100); write("a"); seek(40); read(3);
2007-05-10T10:29:56  <lanius> read always uses the data in the db and the data is not written to the db with write() but with close() or save89
2007-05-10T10:30:20  <xorAxAx> ah, oh well
2007-05-10T10:30:37  <xorAxAx> if that is documneted, its a weird quirk but it would work, yes
2007-05-10T10:30:48  <lanius> essentially it behaves like two objects
2007-05-10T10:33:37  <lanius> ok, acl's again
2007-05-10T10:33:54  <lanius> is the ReadableItem/WritableItem ... approach good?
2007-05-10T10:35:09  <xorAxAx> no
2007-05-10T10:35:24  <xorAxAx> doesnt make any sense IMHO
2007-05-10T10:35:29  <lanius> ;(
2007-05-10T10:35:32  <xorAxAx> how about my idea?
2007-05-10T10:35:32  <lanius> how else
2007-05-10T10:35:38  <lanius> what was your idea again
2007-05-10T10:35:39  <xorAxAx> i have already described it
2007-05-10T10:36:01  <xorAxAx> the revision object just checks on every method the acl, and the item does on getitem as wel
2007-05-10T10:36:15  <xorAxAx> pretty simple and easy
2007-05-10T10:36:45  <xorAxAx> (this is even possible with method decorators, so its clean to implement)
2007-05-10T10:36:56  <lanius> then it needs to know the acl
2007-05-10T10:37:06  <xorAxAx> so?
2007-05-10T10:37:28  <xorAxAx> if we look at the current revision's acl, its a method call of the item
2007-05-10T10:37:29  <lanius> and checking is done on every access -> performace
2007-05-10T10:37:52  <xorAxAx> you can cache that
2007-05-10T10:37:57  <lanius> where to get the user from
2007-05-10T10:38:01  <xorAxAx> performance is not really a reason when it comes to security :)
2007-05-10T10:38:19  <lanius> but if you can gain performance and be as secure as before..
2007-05-10T10:38:36  <xorAxAx> the collection class will supply the user on item creation
2007-05-10T10:39:08  <xorAxAx> umm, better: the request object
2007-05-10T10:39:16  <xorAxAx> which contains the user
2007-05-10T10:39:23  <lanius> and the item take the user from the request?
2007-05-10T10:39:28  <xorAxAx> yes
2007-05-10T10:39:56  <xorAxAx> have you had a look at moin's acls?
2007-05-10T10:40:00  <lanius> so the data object must have a reference back to it's revision back to it's item
2007-05-10T10:40:06  <xorAxAx> why?
2007-05-10T10:40:35  <lanius> how else does the data object checks the acl?
2007-05-10T10:40:36  <xorAxAx> the revision opens it in the correct mode
2007-05-10T10:40:38  <xorAxAx> it doesnt
2007-05-10T10:41:30  <lanius> how does the revision object know the correct acl when the acl from the last revision is always used
2007-05-10T10:42:16  <xorAxAx> i have already said that
2007-05-10T10:42:25  <xorAxAx> 10:37:28 < xorAxAx> if we look at the current revision's acl, its a method call of the item
2007-05-10T10:42:39  <lanius> so the revision must have a reference to it's item
2007-05-10T10:42:46  <xorAxAx> yes, sure
2007-05-10T10:43:06  <lanius> you just said why?
2007-05-10T10:43:14  <xorAxAx> hmm?
2007-05-10T10:43:38  <lanius> and the metadata must have a reference to it's revision
2007-05-10T10:44:01  <lanius> its
2007-05-10T10:44:31  <xorAxAx> i said why to the first part of the sentence
2007-05-10T10:44:46  <lanius> how can i know :D
2007-05-10T10:45:06  <xorAxAx> well, you cant :-)
2007-05-10T10:45:48  <lanius> but the revision must have a reference to its item as well, right
2007-05-10T10:45:53  <lanius> bla
2007-05-10T10:46:01  <lanius> the metadata must have a reference to its revision
2007-05-10T10:46:51  <xorAxAx> now it already has 2 references? :)
2007-05-10T10:47:10  <lanius> metadata to revision, revision to item
2007-05-10T10:47:23  <xorAxAx> i am asking because you said it 2 times :)
2007-05-10T10:47:34  <lanius> i ask because you didn't answered the first time
2007-05-10T10:48:02  <xorAxAx> it wasnt a question :)
2007-05-10T10:48:17  <xorAxAx> but "yes" works here :)
2007-05-10T10:48:33  <lanius> and since we don't have a open method on data anymore i would do the acl checking in data not in revision
2007-05-10T10:49:09  <xorAxAx> ?
2007-05-10T10:49:23  <xorAxAx> i strongly recommend not to do that because it would be a layering violation
2007-05-10T10:49:39  <lanius> you said that the revision opens the data in the right mode, but there is no open
2007-05-10T10:49:39  <xorAxAx> and there is no advantage either IMHO
2007-05-10T10:49:49  <lanius> on that level
2007-05-10T10:49:59  <xorAxAx> you added some mode param to some method
2007-05-10T10:50:24  <lanius> yes, that is the get_data_backend in the StorageBackend
2007-05-10T10:50:36  <lanius> but we are talking about the DataProxy object
2007-05-10T10:50:39  <xorAxAx> just because you removed something without replacement, it doesnt imply that you have to introduce a layering violation instead of fixing it :)
2007-05-10T10:50:51  <lanius> i don't see the layering violation
2007-05-10T10:50:56  <xorAxAx> well, then you have to add the param to the collection
2007-05-10T10:51:03  <xorAxAx> the data backend shouldnt know about items
2007-05-10T10:51:06  <xorAxAx> and acls
2007-05-10T10:51:09  <xorAxAx> thats the violaton :)
2007-05-10T10:51:18  <lanius> then the metadata shouldn't know either
2007-05-10T10:51:32  <xorAxAx> the metadata is on a different layer
2007-05-10T10:51:37  <lanius> it's the same layer
2007-05-10T10:51:39  <xorAxAx> nope
2007-05-10T10:51:48  <xorAxAx> the file class is implemented by the backend
2007-05-10T10:51:51  <xorAxAx> the metadata class isnt
2007-05-10T10:51:58  <lanius> but you have the DataProxy inbetween
2007-05-10T10:52:02  <xorAxAx> ?
2007-05-10T10:52:02  <lanius> which is on the item layer
2007-05-10T10:52:07  <lanius> look at the diagramm
2007-05-10T10:52:14  <lanius> data and metadata are side by side
2007-05-10T10:52:24  <xorAxAx> ah, well
2007-05-10T10:52:31  <xorAxAx> when did you add it? :)
2007-05-10T10:52:42  <lanius> one or two weeks ago?
2007-05-10T10:52:50  <xorAxAx> :-)
2007-05-10T10:53:05  <xorAxAx> ok, then you can do it
2007-05-10T10:53:12  <lanius> :-)
2007-05-10T10:53:21  <xorAxAx> but then you dont need the notion of writable or readable files at all
2007-05-10T10:53:39  <lanius> no in the Data, but in the DataBackend
2007-05-10T10:54:00  <lanius> that's why the open method oved into the StorageBackend
2007-05-10T10:54:01  <xorAxAx> so the databackend should not handle this hybrid file stuff but Data should?
2007-05-10T10:54:07  <lanius> yes
2007-05-10T10:54:10  <xorAxAx> ok, makes sense
2007-05-10T10:54:25  <xorAxAx> (please specify some Data class with these comments in the docstring)
2007-05-10T11:02:40  <dreimark> lanius: can you log in into MM please
2007-05-10T11:03:08  <xorAxAx> yes, would be very nice )
2007-05-10T11:03:17  <lanius> yes
2007-05-10T11:08:26  <dreimark> lanius: I did meet yesterday AndreasSchreiber in cologne
2007-05-10T11:08:53  <lanius> when i call ItemCollection.new_item("test") should the item be created in the backend now or on save()
2007-05-10T11:09:05  <lanius> dreimark: python user group i think
2007-05-10T11:09:26  <lanius> dreimark: i hope he didn't tell too much bad things about me ;)
2007-05-10T11:10:01  <dreimark> dreimark: he told mostly interesting and nice things about how fast you code :)
2007-05-10T11:10:18  <lanius> sounds good :-)
2007-05-10T11:12:35  <dreimark> can we have in metadata something like a checksum or sighnature too ?
2007-05-10T11:12:53  <lanius> wherefor
2007-05-10T11:14:13  <xorAxAx> lanius: i think you would want to create a lazy object that is not saved yet
2007-05-10T11:15:43  <dreimark> lanius: with a checksum we can ommit files saved on different places, we have to only think about something like linking them
2007-05-10T11:16:10  <lanius> xorAxAx: and save() implements all save logic with all calls to the backend then
2007-05-10T11:16:38  <xorAxAx> lanius: how does it know if it has to save a revision?
2007-05-10T11:17:25  <lanius> a revision must only be saved if it was created or removed, that is what the item must know
2007-05-10T11:17:35  <lanius> and on metadata and data it must check a changed flag
2007-05-10T11:18:18  <lanius> or Revision/Metadata/Data has a save method as well, but you get problems if you call save() on metadata when the item wasn't even created
2007-05-10T11:20:13  <lanius> dreimark: i don't understand yet
2007-05-10T11:21:44  <xorAxAx> he seems to be thinking about a distributed hashtable
2007-05-10T11:21:48  <xorAxAx> like googlefs
2007-05-10T11:22:14  <xorAxAx> dreimark: well, the storagebackend can do that internally
2007-05-10T11:23:15  <dreimark> lanius: xorAxAx: I think it should do that particularly for large files
2007-05-10T11:24:16  <xorAxAx> moin is not a file sharing system or cluster storage frontend, its a wiki
2007-05-10T11:28:06  <dreimark> yeah, just an idea if it does not consume much time to implement its nice to have
2007-05-10T11:29:03  <lanius> xorAxAx: what about the save?
2007-05-10T11:29:12  <xorAxAx> lanius: no idea
2007-05-10T11:30:50  <lanius> then i propose my first idea
2007-05-10T11:33:01  <xorAxAx> add it to the page
2007-05-10T11:33:48  <lanius> what about deleting an item
2007-05-10T11:34:21  <xorAxAx> that would be triggered directly, right?
2007-05-10T11:34:32  <lanius> currently yes
2007-05-10T11:35:37  <xorAxAx> please dont over-eagerly remove my comments
2007-05-10T11:35:43  <xorAxAx> you may use (./) though
2007-05-10T11:36:47  <lanius> yes sir :--)
2007-05-10T11:38:14  <xorAxAx> :-)
2007-05-10T11:40:37  <lanius> next one is indexes
2007-05-10T11:40:49  <lanius> what about a simple list in the config
2007-05-10T11:41:25  <xorAxAx> yeah
2007-05-10T11:43:02  <lanius> may i remove your comment then
2007-05-10T11:43:46  <xorAxAx> no
2007-05-10T11:43:52  <lanius> why not
2007-05-10T11:43:54  <xorAxAx> add (./) after it
2007-05-10T11:44:09  <xorAxAx> because i need it as a reminder
2007-05-10T11:44:17  <lanius> ok
2007-05-10T11:44:56  <lanius> what else do we have to discuss
2007-05-10T11:46:14  <xorAxAx> locking :)
2007-05-10T11:46:50  <lanius> yeah, you set the locked attribute in item, which will be directly saved to the backend and on every access the item must check if it is locked
2007-05-10T11:47:25  <xorAxAx> and you want to implement it via meta-data set operations?
2007-05-10T11:47:36  <xorAxAx> so for every lock, you want to serialise the whole metadata?
2007-05-10T11:47:44  <xorAxAx> that will be very expensive :)
2007-05-10T11:48:22  <lanius> per item metadata and add a method for just getting one metadata key :D
2007-05-10T11:48:56  <ThomasWaldmann> lanius: moin :) i added a comment to the acl stuff.
2007-05-10T11:49:08  <lanius> moin Thomas
2007-05-10T11:49:51  <xorAxAx> lanius: so you dont want any read locks?
2007-05-10T11:50:02  <xorAxAx> lanius: what if somebody is reading and you are setting the write lock
2007-05-10T11:50:14  <xorAxAx> then you modify the file and the reader might have trouble
2007-05-10T11:50:57  <xorAxAx> and what if 2 requests modify metadata at the same time?
2007-05-10T11:51:17  <lanius> how is it currently done
2007-05-10T11:51:24  <xorAxAx> both read, then lock, then write, then unlock. if the last 3 steps are sequential, nobody will benefit from the lock and you will have data loss
2007-05-10T11:51:38  <xorAxAx> basically, metadata and data is immutable and for caching files, there are locks
2007-05-10T11:52:35  <xorAxAx> (and for some cache files, read locks are disabled because corruption while reading would be no serious harm)
2007-05-10T11:55:32  <lanius> what's your idea for the new storage?
2007-05-10T11:55:53  <xorAxAx> having some concept of immutable and mutable data would be nice
2007-05-10T11:56:06  <xorAxAx> or you can dump some caching stuff into the file system
2007-05-10T11:56:18  <lanius> so if you get the item from the collection it will be mutable or immutable?
2007-05-10T11:56:57  <xorAxAx> well, some parts of it have to be mutable
2007-05-10T11:57:11  <xorAxAx> i added ~ 10 new questions to my comments section and have to go now, c u
2007-05-10T11:57:15  <xorAxAx> i will be back in the afternoon
2007-05-10T11:57:19  <xorAxAx> ;-)
2007-05-10T11:57:21  <lanius> ^^ me not, cu
2007-05-10T13:15:18  <ThomasWaldmann> vpv: http://code.google.com/soc/fedora/about.html where is your project?
2007-05-10T13:27:07  <ThomasWaldmann> vpv: btw, you could create a homepage on mm site for you
2007-05-10T13:27:33  <ThomasWaldmann> bbl
2007-05-10T14:49:09  <vpv> ThomasWaldmann: It's not a Google Summer of Code project, the Finnish Summercode is a different project, I'll give you the URL
2007-05-10T14:49:36  <vpv> ThomasWaldmann: http://www.coss.fi/web/coss/developers/summercode
2007-05-10T14:49:55  <vpv> ThomasWaldmann: and http://www.coss.fi/web/coss/developers/summercode/2007
2007-05-10T15:18:49  <ThomasWaldmann> ah, ok, I confused that X)
2007-05-10T15:21:57  <ThomasWaldmann> maybe put those urls on your homepage, too :)
2007-05-10T15:23:43  <ThomasWaldmann> vpv: btw, if one would do one man-page wiki per (fedora) release, one could use sisterpages to click through them
2007-05-10T15:24:04  <ThomasWaldmann> otoh, there is no diff between sisterpages
2007-05-10T15:26:13  <ThomasWaldmann> also, you have to think about the format you want to put into the wiki
2007-05-10T15:29:20  <ThomasWaldmann> *roff? (not nice to edit) docbook? other?
2007-05-10T15:30:23  <vpv> I think the idea would be to put all the man pages in the same wiki. I was thinking about child pages (not sure what they're called in MM) so that you would have something like foowiki.bar/fc6/<section number here maybe>/printf or...
2007-05-10T15:31:27  <vpv> Docbook might be one option, I don't know it that well myself, but I'm ready to learn. IIRC Karsten also mentioned TeX as one possibility
2007-05-10T15:32:16  <ThomasWaldmann> for the *TeX stuff you can talk to johill, he can tell you the security implications :)
2007-05-10T15:33:33  <ThomasWaldmann> vpv: in soc 2006 mikko v. worked on some docbook stuff, maybe have a look at that
2007-05-10T15:33:55  <ThomasWaldmann> (that's the stuff where the docs still need to get improved)
2007-05-10T15:34:17  <ThomasWaldmann> http://hg.thinkmo.de/moin/1.6-docbook-mvirkkil/
2007-05-10T15:35:24  <ThomasWaldmann> (maybe search for mvirkkil to see his changesets)
2007-05-10T15:35:55  <vpv> yeah, I'm aware of what he did, but I don't know the code, I wasn't yet part of Fedora translation or documentation a year ago.
2007-05-10T15:37:37  <ThomasWaldmann> the problem is that I personally don't use docbook, so I can't tell much about it
2007-05-10T15:38:33  <vpv> I bet Karsten and the others at Fedora Docs could tell me about it, since they use it a lot
2007-05-10T15:39:04  <ThomasWaldmann> sure :)
2007-05-10T15:40:28  <ThomasWaldmann> maybe also check if the versioning you need is just linear or whether you need forks
2007-05-10T15:41:24  <ThomasWaldmann> if it is just linear, like revision 1 of some man page being fc5 and revision 10 being fc6, we maybe could use some sort of tagging
2007-05-10T15:41:41  <ThomasWaldmann> (and just a single wiki page)
2007-05-10T15:41:54  <ThomasWaldmann> diff for free :)
2007-05-10T15:42:35  <ThomasWaldmann> of course that does not work if you need to work on different branches of something at the same time
2007-05-10T15:44:52  <vpv> I think I need some sort of branching, because the system should be able to support something like having enterprise linux and fedora man pages at the same time and those might be different
2007-05-10T17:22:19  <xorAxAx> the fedora ones would end with "if you want to continue reading, please buy your RHEL license" :)
2007-05-10T18:17:53  <neagulm> hello
2007-05-10T18:31:19  <ThomasWaldmann> moin neagulm
2007-05-10T19:22:29  <grzywacz> moin
2007-05-10T19:23:09  <xorAxAx> hi grzywacz
2007-05-10T19:23:24  <grzywacz> Hello HMRE!
2007-05-10T19:23:25  <grzywacz> ;-)
2007-05-10T19:23:32  <xorAxAx> :-)
2007-05-10T21:03:52  <dreimark_> moin
2007-05-10T21:04:24  <dreimark_> hi neagulm
2007-05-10T21:04:32  <neagulm> hy dreimark_
2007-05-10T21:04:46  <neagulm> dreimark_, i was missing a few days :(
2007-05-10T21:04:57  <neagulm> dreimark_, exams
2007-05-10T21:05:12  <flowhase> hi there
2007-05-10T21:05:22  <dreimark_> I recognized, ahh ok, and whats your feeling about
2007-05-10T21:05:28  <flowhase> sorry for being so quiet the last days
2007-05-10T21:05:31  <dreimark_> hi flowhase
2007-05-10T21:05:41  <flowhase> i had some personal issues to take care of
2007-05-10T21:06:08  <dreimark_> neagulm: did you get some more respond by skip?
2007-05-10T21:06:45  <neagulm> dreimark_, no
2007-05-10T21:08:17  <neagulm> dreimark_, until Tuesday i hope to finish PoC v1.9
2007-05-10T21:08:37  <neagulm> dreimark_, a fully functional patch to moinmoin
2007-05-10T21:09:27  <dreimark_> fully functional sounds amazing what exactly is it
2007-05-10T21:10:16  <dreimark_> already the whole controls?
2007-05-10T21:10:29  <dreimark_> neagulm: ?
2007-05-10T21:10:42  <xorAxAx> hi flowhase
2007-05-10T21:10:50  <xorAxAx> flowhase: did you put the log into the wiki (of your discussion)?
2007-05-10T21:13:46  <neagulm> dreimark_, no traceback's, work 'almost' as I want. :D
2007-05-10T21:14:12  <dreimark_> neagulm: :)
2007-05-10T21:15:39  <xorAxAx> flowhase: or at least the results/ideas
2007-05-10T21:16:37  <flowhase> xorAxAx: i will do it this evening
2007-05-10T21:16:45  <xorAxAx> cool
2007-05-10T21:16:51  <flowhase> just want to relax for half an hour or so :)
2007-05-10T21:17:26  <flowhase> life was quite bitchy and i have to reconfigure my brain a little bit
2007-05-10T21:21:24  <dreimark_> neagulm: I think we should invite Skip to irc again after you have finished finish PoC v1.9
2007-05-10T21:22:05  <xorAxAx> PoCß
2007-05-10T21:22:07  <xorAxAx> ...?
2007-05-10T21:23:06  <dreimark_> xorAxAx: ProofOfConcept
2007-05-10T21:25:00  <xorAxAx> ah
2007-05-10T21:26:08  <dreimark_> xorAxAx: see http://hg.thinkmo.de/moin/1.7-classify-mneagul/
2007-05-10T21:28:42  <neagulm> dreimark_, I agree with inviting Skip :)
2007-05-10T21:34:33  <dreimark_> neagulm:ok, it would be nice if you can summarize a bit on the wiki page what all 1.9 is about in comparison to the last one
2007-05-10T21:36:42  <neagulm> dreimark_, it will be done :)
2007-05-10T21:36:47  <neagulm> dreimark_, brb
2007-05-10T21:38:19  <dreimark_> xorAxAx, ThomasWaldmann: can we change Page.exists in that way that it checks for pagedir/current instead of only pagedir?
2007-05-10T21:38:50  <xorAxAx> dreimark_: ? it already does
2007-05-10T21:39:02  <xorAxAx> dreimark_: otherwise you may have seen a caching bug
2007-05-10T21:40:06  <dreimark_> in 1.6 it doesn't,  I do have empty dirs e.g. MoinMoinBugs/RenamePageThatWasPreviouslyPreviewed
2007-05-10T21:41:22  <xorAxAx> dreimark_: how is that rename issue related to exists?
2007-05-10T21:41:32  <dreimark_> xorAxAx: foo is created and its an empty dir now and I cant rename to that
2007-05-10T21:41:41  <xorAxAx> thats an orthogonal issue
2007-05-10T21:49:19  <dreimark_> on the other hand if only that empty dir exist it could be removed by canceling?
2007-05-10T21:53:56  <xorAxAx> well, that wouldnt be enough
2007-05-10T22:13:30  <dreimark_> xorAxAx: what else?
2007-05-10T22:14:40  <dreimark_> ah ok locking could be a problem
2007-05-10T22:14:48  <xorAxAx> no
2007-05-10T22:14:58  <xorAxAx> people who dont click cancel
2007-05-10T22:23:03  <neagulm> dreimark_, I need to leave
2007-05-10T22:23:20  <neagulm> dreimark_, I will be back tomorrow
2007-05-10T22:23:31  <dreimark_> neagulm: ok, good night
2007-05-10T22:24:19  <neagulm> dreimark_, good night
2007-05-10T22:34:17  <dreimark_> good night