2008-03-29T00:13:56 <xorAxAx> LOL, leslie is encouraging screen scraping
2008-03-29T00:14:30 <dennda> screen scraping?
2008-03-29T00:14:35 <dennda> with a knife?
2008-03-29T00:14:46 <xorAxAx> johill: having edit locking parallely to items sounds like a bad thing imho
2008-03-29T00:15:13 <xorAxAx> dennda: and sometimes even beautiful soup or html5 :)
2008-03-29T00:17:22 <dennda> hm?
2008-03-29T00:18:29 <xorAxAx> checkout both python libs
2008-03-29T00:18:34 <johill> dennda: screen scraping (in this context) means to download a website with a script and try to parse it
2008-03-29T00:18:46 <johill> xorAxAx: let me try to figure out why I didn't like it
2008-03-29T00:19:28 <dennda> I know beautiful soup
2008-03-29T00:19:53 <dennda> I did that once to download all icanhascheezburger.com cats
2008-03-29T00:20:02 <johill> hahaha
2008-03-29T00:21:36 <xorAxAx> dennda: see, thats screen scraping
2008-03-29T00:24:19 <dennda> and I didn't even know :)
2008-03-29T00:24:47 <xorAxAx> hehe
2008-03-29T00:29:29 <xorAxAx> johill: your acl thoughts are too complex
2008-03-29T00:29:36 <xorAxAx> observe the following nicety:
2008-03-29T00:30:01 <xorAxAx> legacy code will never run into an exception thrown by the acl checking code of the new storage stuff
2008-03-29T00:30:08 <xorAxAx> because it will check for permissions first
2008-03-29T00:37:22 <johill> yes and?
2008-03-29T00:37:47 <johill> new code doesn't want to have to do that
2008-03-29T00:38:02 <johill> and as I explain, it's not possible to avoid an up-front deleted check in the current design
2008-03-29T00:39:12 <xorAxAx> johill: no, this stuff doesnt make much sense: "That also allows converting code"
2008-03-29T00:39:29 <johill> ah ok, yeah
2008-03-29T00:40:19 <johill> that's not really the main point anyway
2008-03-29T00:40:57 <xorAxAx> well, one question is whethr wrapping is sane. it is, imho, iff nobody really circumvents the wrapper
2008-03-29T00:41:03 <xorAxAx> and the wrapper is sound
2008-03-29T00:41:10 <xorAxAx> because people will start to rely on it
2008-03-29T00:41:56 <johill> I'm not contesting that
2008-03-29T00:42:41 <johill> but it's not possible to implement delete acl checking in the current wrapper (ItemCollection) easily since revisions might be left with full data but w/o the deleted metadata since only that actually raised the exception
2008-03-29T00:43:19 <xorAxAx> "required to lock the revision even against concurrent read access" - why would tht be necessary? :)
2008-03-29T00:43:50 <johill> think first?
2008-03-29T00:43:51 <johill> :P
2008-03-29T00:44:07 <johill> because if item data is changed you cannot guarantee consistent data without that
2008-03-29T00:44:49 <xorAxAx> but item data doesnt change
2008-03-29T00:44:58 <johill> e.g. if you implement metadata as a table |id|revisionid|key|value|
2008-03-29T00:45:02 <johill> metadata
2008-03-29T00:45:59 <xorAxAx> so you are not talking about data?
2008-03-29T00:46:15 <johill> it doesn't really matter much whether we want to change data or metadata
2008-03-29T00:46:22 <xorAxAx> no :)
2008-03-29T00:46:25 <johill> the actual use for user items is probably metadata
2008-03-29T00:46:37 <xorAxAx> data is assumed to be read only in our model for practical reasons
2008-03-29T00:46:50 <johill> what I'm saying is that it's not possible to guarantee a consistent view of data/metadata if it is allowed to change
2008-03-29T00:47:02 <johill> and in a user storage it does change unless you create revisions all the time
2008-03-29T00:47:13 <xorAxAx> well
2008-03-29T00:47:26 <xorAxAx> where is the writelock that is the pendant for this readlock?
2008-03-29T00:47:34 <xorAxAx> that you sketch here
2008-03-29T00:47:53 <johill> whenever you update the user item, i.e. the user goes and changes his email or whatever
2008-03-29T00:49:20 <xorAxAx> johill: where == on which level
2008-03-29T00:50:30 <johill> well I would put it both into the backend level since my revision objects are just shadows and only committed when you .save() the item object they belong to
2008-03-29T00:50:59 <xorAxAx> the backend only sees single metadata writes currently
2008-03-29T00:51:10 <johill> exactly the problem
2008-03-29T00:51:34 <xorAxAx> well, your lockfree file discusses problems shortly but is not very concrete on the design side of things :)
2008-03-29T00:51:54 <johill> you can't create consistent views with that unless you throw in locking at a higher level, and even then renames are still not possible to do
2008-03-29T00:52:11 <xorAxAx> well, how about dirty reads?
2008-03-29T00:52:27 <xorAxAx> you read metadata, somebody else changes it, you pass it to the backend and overwrite his data
2008-03-29T00:52:40 <xorAxAx> that doesnt really conflict with your ideas
2008-03-29T00:52:53 <johill> nope, doesn't
2008-03-29T00:53:06 <xorAxAx> in the current state, it should be guarded by early taken write locks
2008-03-29T00:53:22 <xorAxAx> as you can find documented in the discussions about this topic :)
2008-03-29T00:53:32 <johill> yeah it really plays together, a write lock you will still need for actually modifying an item
2008-03-29T00:53:36 <xorAxAx> not sure if the code actually does that, i wasnt the mentor
2008-03-29T00:53:54 <johill> but you don't actually need a higher level read lock if you can guarantee consistent views
2008-03-29T00:54:43 <xorAxAx> so i wonder how you solve this "lost update" (not "dirty read") problem, good night :)
2008-03-29T00:55:06 <johill> it's not a problem you can solve at the backend level
2008-03-29T00:55:16 <xorAxAx> wellll
2008-03-29T00:55:16 <johill> you still solve it with a write lock at higher levels
2008-03-29T00:55:25 <johill> for user storage anyway
2008-03-29T00:55:26 <xorAxAx> then please rename the file %-)
2008-03-29T00:55:31 <xorAxAx> gn^2
2008-03-29T00:56:10 <johill> night
2008-03-29T00:56:12 <johill> whatever
2008-03-29T00:56:29 <johill> you always get so hung up about details rather than trying to understand the big picture first
2008-03-29T00:57:09 <johill> besides, for *page* storage you really don't need locks
2008-03-29T00:59:42 <johill> user storage with mutable metadata really was an after-thought
2008-03-29T10:07:10 <zenhase> moin
2008-03-29T10:21:49 <gizmach> zenhase: :)
2008-03-29T10:23:34 <zenhase> gizmach: hi there. how's your application doing?
2008-03-29T10:33:02 <gizmach> zenhase: I'm writing it right now, we'll se how will it be because I'm a moin newbie :)
2008-03-29T10:33:39 <xorAxAx> johill: indeed
2008-03-29T10:33:46 <xorAxAx> johill: well, thats why i went to bed :)
2008-03-29T10:37:52 <gizmach> I plan to upload the first version on my website
2008-03-29T10:37:57 <gizmach> xorAxAx: morning
2008-03-29T10:38:03 <xorAxAx> good morning gizmach :)
2008-03-29T10:46:47 <zenhase> good moin, xorAxAx
2008-03-29T10:47:45 <johill> xorAxAx: do you really care more about the filename than the contents? it seems that way to me :/
2008-03-29T10:50:55 <xorAxAx> zenhase: bzzzt :)
2008-03-29T10:51:00 <xorAxAx> johill: no
2008-03-29T10:51:12 <zenhase> bzzt? oO
2008-03-29T10:51:22 <xorAxAx> zenhase: sound for wrong language usage
2008-03-29T10:51:30 <xorAxAx> johill: who will implement the transactional locking then?
2008-03-29T10:51:47 <xorAxAx> johill: every user of the backend because you will remove item collection?
2008-03-29T10:51:48 <zenhase> xorAxAx: i don't see any 'wrong language usage' :P
2008-03-29T10:51:58 <xorAxAx> zenhase: :)
2008-03-29T10:52:04 <zenhase> it's a fun and creative combination
2008-03-29T10:52:09 <zenhase> language mashups :P
2008-03-29T10:52:20 <johill> xorAxAx: I dont' see how that's any different whether you call Item.lock() on an item you got from the backend or from the ItemCollection
2008-03-29T10:52:36 <johill> xorAxAx: the code for that could remain to live in the Item class since that's something backend items would inherit from
2008-03-29T10:52:50 <xorAxAx> johill: inherit?!
2008-03-29T10:54:20 <xorAxAx> (seems wrong to me, it would completly give up decoupling)
2008-03-29T10:54:23 <johill> what's wrong with actually using the OO properties of python?
2008-03-29T10:55:07 <xorAxAx> hmm
2008-03-29T10:55:10 <johill> fine then keep an ItemCollection that has Items that keep backend-item-objects in them
2008-03-29T10:55:14 <johill> I don't see the difference
2008-03-29T10:55:21 <johill> I don't see why you always think layering needs to be absolute
2008-03-29T10:55:22 <xorAxAx> i didnt say that this is so much nicer :)
2008-03-29T10:55:47 <xorAxAx> well, i guess inheriting works out
2008-03-29T10:55:52 <johill> it's not like we want RPC between teh different code layers so what does it matter?
2008-03-29T10:55:56 <xorAxAx> if the apis are well documented
2008-03-29T10:56:08 <xorAxAx> johill: in python, thats not a reason to choose either design :)
2008-03-29T10:56:15 <xorAxAx> (the rpc one)
2008-03-29T10:57:18 <johill> even that is true :)
2008-03-29T10:57:33 <johill> I think the Item class can pretty much stay as is
2008-03-29T10:57:40 <johill> it just needs to call the backend with itself rather than the item name
2008-03-29T10:57:53 <johill> which would already solve most of my complaints if the backend was also responsible for instantiating it
2008-03-29T10:58:29 <johill> because then e.g. the sql backend could store the unique page ID in the item class and use that all over instead of doing name lookups again and again, thereby already fixing the rename race
2008-03-29T10:58:32 <xorAxAx> hmm, somehow elegant
2008-03-29T10:59:05 <johill> that's my biggest complaint really that the name is always passed in again and again and you need to do lots of work in non-fs backends to accomodate that
2008-03-29T10:59:36 <johill> if the backend was to allocate the Item it still wouldn't need much different methods since the base Item class methods would just call the backend pretty much as it is now
2008-03-29T10:59:53 <johill> of course, I'd like to see different update semantics too, but most of that is just icing on top
2008-03-29T11:06:55 <xorAxAx> johill: where would the commit tracking related code reside?
2008-03-29T11:07:05 <xorAxAx> i.e. the code that does the right thing on save()?
2008-03-29T11:20:50 <johill> that would have to be in the backend mostly
2008-03-29T11:20:56 <johill> the Item class could help with caching the values
2008-03-29T11:21:12 <johill> but .save would just be a backend method that gets the item with the cached revision values etc
2008-03-29T11:25:00 <kikka> moin
2008-03-29T11:25:30 <xorAxAx> johill: umm
2008-03-29T11:25:42 <xorAxAx> johill: i guess you still need modified flags etc.
2008-03-29T11:25:43 <johill> maybe it could have a list of new revision objects, or maybe revision objects need to be created through the backend as well rather than in the item object
2008-03-29T11:26:00 <xorAxAx> we also had this design in mind
2008-03-29T11:26:09 <xorAxAx> if you want to know why it wasnt chosen, see the logs :-)
2008-03-29T11:26:16 * xorAxAx doesnt remember
2008-03-29T11:26:19 <johill> yes you will need to keep track of what was changed
2008-03-29T11:27:12 <johill> you know what, I really don't care why it wasn't chosen, I don't care about my design much either, but the design flaws in the current design are big enough to make it useless
2008-03-29T11:27:40 <johill> you can only make it work with read/write locking around every access, and that's far too expensive
2008-03-29T11:28:23 <johill> and seriously, I have no idea how I would ever find that in logs of mostly useless conversations
2008-03-29T11:28:54 <dreimark> without reading those completly
2008-03-29T11:29:18 <xorAxAx> johill: well, thats the art of searching :)
2008-03-29T11:29:28 <johill> so I reiterate: pointing to logs for design decisiosn is completely useless and actually telling people that the design decisions are not properly documented
2008-03-29T11:29:29 <xorAxAx> well, its better, indeed
2008-03-29T11:29:46 <xorAxAx> johill: yes, i dont care. i didnt do the final ones :)
2008-03-29T11:30:07 <johill> I don't see why you keep mostly defending them then :)
2008-03-29T11:30:18 <zenhase> moin kikka, moin dreimark
2008-03-29T11:30:38 <kikka> heyho zenhase! :)
2008-03-29T11:30:43 <johill> anyhow, I don't see tracking what was changed as a problem
2008-03-29T11:31:01 <johill> it's fairly simple to do since everything goes through two or three classes
2008-03-29T11:31:30 <johill> and for page storage, all changes that can actually be done are creating a new revision
2008-03-29T11:31:36 <johill> (with data and metadata)
2008-03-29T11:31:43 <xorAxAx> johill: because i dont see any strong point not to search them :)
2008-03-29T11:32:00 <johill> (except for the item-metadata thing)
2008-03-29T11:32:26 <johill> xorAxAx: sure, it's an utter waste of my time, I know the design is broken so I don't need to look for the decisions that lead there :)
2008-03-29T11:32:35 <dreimark> hi zenhase, are you inspired and do you have choosen something from the user interface enhancement?
2008-03-29T11:32:55 <xorAxAx> johill: i dont think that those are primarily documented there but maybe you will find good reasons not to go down this path :)
2008-03-29T11:33:05 <xorAxAx> johill: of course it might be easier to ask heinrich :)
2008-03-29T11:33:07 <xorAxAx> lanius:
2008-03-29T11:33:14 <zenhase> dreimark: uhm, i didn't look very closely on the UI tasks
2008-03-29T11:33:23 <zenhase> dreimark: actually i had 2 other ideas
2008-03-29T11:33:36 <dreimark> cool
2008-03-29T11:33:46 <zenhase> one is the WSGI refactoring
2008-03-29T11:34:22 <johill> xorAxAx: the only thing I managed to extract from him so far about that is that one of the foremost design decisions he made is that the backend must be trivial to implement, which he feels is not given with such a design
2008-03-29T11:34:37 <xorAxAx> hmm
2008-03-29T11:34:42 <zenhase> the other one would be ... uhm ... retry send_page-refactoring?
2008-03-29T11:34:50 <johill> which is true, but actually enables doing useful work with the backend
2008-03-29T11:35:10 <dreimark> zenhase: yeah, that's an important one too
2008-03-29T11:35:17 <zenhase> i thought so
2008-03-29T11:35:27 <johill> zenhase: wsgi would be cool, and it could help a lot with tests too
2008-03-29T11:35:32 <zenhase> johill: yeah
2008-03-29T11:35:40 <zenhase> johill: that's what i was thinking about
2008-03-29T11:36:11 <zenhase> johill: reduce the request-specific code to one code-base (only wsgi-request), and some adapters for wsgi as servers
2008-03-29T11:36:20 <ThomasWaldmann> moin
2008-03-29T11:36:32 <johill> zenhase: I actually tried doing a wsgi-standalone server and that wasn't too hard
2008-03-29T11:36:45 <zenhase> johill: i know, it's really not that hard
2008-03-29T11:37:01 <zenhase> johill: not everything has to be fullblown wsgi like Paste ;)
2008-03-29T11:37:03 <johill> still lots of work though if you want to keep all the debug stuff working
2008-03-29T11:37:16 <johill> like pycallgraph etc.
2008-03-29T11:37:29 <zenhase> yeah
2008-03-29T11:37:50 <zenhase> and in the longterm some stuff can be moved down to the middleware layers
2008-03-29T11:37:52 <johill> and I haven't managed to fit in the wsgi static file serve thing yet
2008-03-29T11:38:05 <zenhase> some of the auth-stuff for example
2008-03-29T11:38:11 <zenhase> hmm
2008-03-29T11:38:33 <zenhase> well, that would be quite some stuff to do for WSGI
2008-03-29T11:38:34 <johill> yeah, it could be, but I'm generally not very happy with the wsgi auth solutions, they're too far from the application imho
2008-03-29T11:38:59 <johill> oh don't get me wrong, it's certainly doable :) just not something I do in a few days ;)
2008-03-29T11:39:00 <zenhase> and well ... as dreimark said: send_page is a helluva task too :o
2008-03-29T11:40:14 <zenhase> it would be nice to see success where i stuck and finally failed last year
2008-03-29T11:40:18 <johill> xorAxAx: I stopped asking at that point though since I felt that designing something around how easy the implementation is rather than considering thread-safety issues up-front was so wrong I didn't watn to know ;)
2008-03-29T11:40:20 <dreimark> (that was from ThomasWaldmann)
2008-03-29T11:40:24 <zenhase> s/stuck/got stuck
2008-03-29T11:40:43 <xorAxAx> johill: i have never perceived that argument in the discussions we had with him last year
2008-03-29T11:41:01 <johill> odd, it came up one of the first things when I started asking
2008-03-29T11:41:27 <zenhase> you're talking about storage layer again, right?
2008-03-29T11:41:32 <johill> yeah
2008-03-29T11:41:38 <zenhase> ok, just got confused for a second
2008-03-29T11:41:57 <johill> anyway, it would be good to have the design assumptions documented regardless of whether I tear it apart or not, and I'm not going to dig through the logs to do that
2008-03-29T11:42:03 <xorAxAx> zenhase: "Just because you're paranoid doesn't mean they aren't after you" :)
2008-03-29T11:42:19 <xorAxAx> johill: good idea
2008-03-29T11:42:29 <zenhase> well, i will browse some code today and then will write applications and the rough outlines for WSGI and send_page
2008-03-29T11:42:38 <xorAxAx> maybe lanius feels motivated to do so and maybe compare your new ideas
2008-03-29T11:42:41 <zenhase> i give it a shot
2008-03-29T11:42:49 <johill> xorAxAx: I actually discussed this during the design phase too, see the diagram on my wiki homepage
2008-03-29T11:42:52 <dreimark> zenhase: :)
2008-03-29T11:43:02 <xorAxAx> johill: hehe
2008-03-29T11:43:21 <johill> however, at the time, I wasn't aware how his design would fail
2008-03-29T11:43:28 <johill> so I didn't push further after he rejected it
2008-03-29T11:43:30 <zenhase> if send_page get's finished i could start on my theme-code again after soc
2008-03-29T11:43:47 <zenhase> uha ... s/get's/gets
2008-03-29T11:43:54 <zenhase> *shiver*
2008-03-29T11:44:49 <dreimark> I think we should ask students this time to summarize decissions done during an irc session
2008-03-29T11:45:48 <dreimark> zenhase: yeah that would be a nice task then for 09 again
2008-03-29T11:46:10 <johill> 2007-05-04T17:43:14 <johill> I think the main difference is that with lanius's design everything is basically keyed off the page name etc, while with my design a lot more things are implicit by the class
2008-03-29T11:46:36 <johill> I think I just said the same today too ;)
2008-03-29T11:46:50 <xorAxAx> lol
2008-03-29T11:47:15 <dreimark> dejavu
2008-03-29T11:47:28 <johill> but now I actually know why it makes sense to not use the page name all the time
2008-03-29T11:47:38 <zenhase> dreimark: doesn't have to be explicitly a soc-task imo, but will see
2008-03-29T11:49:36 <johill> a bit later that day you got into discussing thread/process-safety issues
2008-03-29T11:49:47 <johill> xorAxAx: and you suggested the item-level lock ;)
2008-03-29T11:50:12 <xorAxAx> johill: well, its necessary, isnt it?
2008-03-29T11:50:14 <johill> and lanius confuses it with the edit-lock, at least at first, apparently
2008-03-29T11:50:31 <xorAxAx> to gain transactional safety against lost updates for stuff
2008-03-29T11:50:34 <dreimark> johill: btw. can the redisgn solve to know where a pagelink was used to an existing page and can that be renamed then too?
2008-03-29T11:50:48 <xorAxAx> johill: :)
2008-03-29T11:50:48 <johill> dreimark: let me hold back that question for a second
2008-03-29T11:51:04 <johill> xorAxAx: no, you only need lost update safety for stuff where you actually *change* data
2008-03-29T11:51:19 <johill> xorAxAx: my design assumes that revision data and metadata is immutable once saved
2008-03-29T11:51:32 <johill> in which case the only thing you need to know is whether that revision you're trying to create doesn't exist yet
2008-03-29T11:51:40 <johill> scroll all the way down in LOCKFREE to the example code
2008-03-29T11:52:17 <xorAxAx> ok
2008-03-29T11:52:32 <johill> dreimark: no, it doesn't explicitly solve that, and it cannot really, because then you'd have to transform the stored data into something that contained 'unique IDs' instead of page links
2008-03-29T11:53:04 <johill> dreimark: unless you want to update all pages that link to one, but that would have to create a new revision for each page with the associated edit-lock/conflict problems
2008-03-29T11:53:34 <johill> dreimark: also, I'm not sure that is worth solving, the user explicitly linked to a page on something by its title, so if it stops existing then that is some information in itself
2008-03-29T11:55:12 <johill> xorAxAx: the lost update problem is really only relevant for user storage where an existing item/revision is modified
2008-03-29T11:55:33 <xorAxAx> johill: or item-wide metadata
2008-03-29T11:55:37 <johill> xorAxAx: there, the higher levels will have to have a lock to guard against that, but they can still work with concurrent readers if .save can do it
2008-03-29T11:55:51 <johill> yes, anything that isn't immutable
2008-03-29T11:56:00 <xorAxAx> yep
2008-03-29T11:56:23 <johill> I remember now why I don't like item-wide metadata
2008-03-29T11:56:38 <johill> because then you can easily end up with pages that have no revisions, and there's no easy way to remove them again once all metadata keys are removed from them
2008-03-29T11:56:59 <johill> say you start editing, you need to save an item that is empty except for its metadata
2008-03-29T11:57:14 <dreimark> I you refactor code e.g. bicycle repair man knows what a class, a method or a parameter is.
2008-03-29T11:57:20 <dreimark> johill:
2008-03-29T11:57:28 <dreimark> s/I/If/
2008-03-29T11:57:45 <xorAxAx> johill: well, not necessarily
2008-03-29T11:57:50 <xorAxAx> thats kind of orthogonal
2008-03-29T11:57:51 <johill> now when you cancel the edit you want to delete that item again, but that's extremely hard to do properly if somebody else saves the item
2008-03-29T11:58:05 <johill> I mean, to protect against problems there
2008-03-29T11:58:06 <xorAxAx> it depends on wheether you actually need to store metadata before the user does the save
2008-03-29T11:58:13 <xorAxAx> and i dont see why that would be necessary
2008-03-29T11:58:21 <gizmo> back, my father was playing with electricity at home so I got disconected
2008-03-29T11:58:22 <johill> (see BUG 4 in my BUGS file about what can happen)
2008-03-29T11:58:26 <johill> dreimark: ??
2008-03-29T11:58:33 <johill> xorAxAx: edit-locks are metadata
2008-03-29T11:58:48 <johill> xorAxAx: so you need to keep them for editing conflicts
2008-03-29T11:58:51 <johill> on new pages
2008-03-29T11:58:56 <xorAxAx> johill: well, yes. i still think that they are not necessary for a new page
2008-03-29T11:59:01 <xorAxAx> johill: but nevertheless there is this bug
2008-03-29T11:59:05 <xorAxAx> which is orthogonal :)
2008-03-29T11:59:17 <johill> yeah the bug is orthogonal I just found it because I was thinking about this issue
2008-03-29T11:59:35 <johill> yes you can say that edit locks aren't necessary for new items
2008-03-29T11:59:47 <CIA-39> Johannes Berg <johannes AT sipsolutions DOT net> default * 3770:117e19b424f6 1.7-storage-hwendel/LOCKFREE: fix indentation in example code
2008-03-29T11:59:48 <johill> but then you'd better guarantee that no metadata will ever be saved to an empty item
2008-03-29T11:59:49 <xorAxAx> so, somebody fixing the bug shouldnt introduce the edit-lock like it had been in 1.3
2008-03-29T12:00:01 <xorAxAx> johill: makes some kind of sense, yes
2008-03-29T12:00:08 <johill> or rather, that an item w/o revisions cannot be crated
2008-03-29T12:00:10 <johill> created
2008-03-29T12:00:12 <ThomasWaldmann> johill: + if not str(err).endswith('wikiconfig_local'): # is that checking for the ImportError?
2008-03-29T12:00:25 <johill> in lanius's design, however, the empty item is always around from the start
2008-03-29T12:00:44 <johill> ThomasWaldmann: yeah that's a bit sucky, but it tries to catch only the import error for the wikiconfig_local file, not anything that might have happened in that file
2008-03-29T12:01:12 <xorAxAx> brb
2008-03-29T12:01:25 <ThomasWaldmann> maybe the check then should include ImportError somehow?
2008-03-29T12:01:41 <johill> ThomasWaldmann: so if you have 'import foo' in your local conf it will still show up
2008-03-29T12:01:51 <johill> ThomasWaldmann: eh, it runs in an except ImportError, err: block
2008-03-29T12:02:32 <johill> brb 2
2008-03-29T12:03:29 <ThomasWaldmann> ah, ok, overlooked that
2008-03-29T12:05:06 <johill> :)
2008-03-29T12:05:15 <johill> I agree with you, I don't like it much
2008-03-29T12:06:24 <johill> didn't see any other way though since it's created only with that string in it
2008-03-29T12:07:56 <dreimark> johill: currently we go to make mimetypes as items. we have to destinguish for rename already that we don't change the mimetype by this action. I think we should make it possible to change the content of a page too if one of the items or the page itself was renamed
2008-03-29T12:09:04 <johill> dreimark: hm?
2008-03-29T12:09:15 <johill> dreimark: a mimetype is metadata of an item so it just follows the rename
2008-03-29T12:09:27 <johill> dreimark: a revision of that item should be immutable, if you chhange the item you create a new revision
2008-03-29T12:09:34 <dreimark> and can't displayed then
2008-03-29T12:09:39 <johill> why not?
2008-03-29T12:09:43 <dreimark> e.g. imge.png -> image.ogg
2008-03-29T12:09:49 <johill> ah
2008-03-29T12:09:51 * ThomasWaldmann will migrate the last production 1.5 wiki on our farm now
2008-03-29T12:09:53 <johill> no, mimetype is revision metadata
2008-03-29T12:10:04 <johill> sorry for not being precise
2008-03-29T12:15:17 <dreimark> in the sql database I guess each item has a uniq id or?
2008-03-29T12:15:41 <johill> everything pretty much has, yeah
2008-03-29T12:15:58 <johill> but that's more of an implementation detail that facilitates things
2008-03-29T12:16:06 <johill> I'm not sure we should let that escape the backend layer
2008-03-29T12:16:27 <johill> not without good design, anyway
2008-03-29T12:17:17 <johill> it might be good to have unique IDs, and it might also be good to have the backend assign them (so you don't need to look up metadata to get the ID)
2008-03-29T12:18:02 <dennda> flyer und cds sollte man immer genug haben
2008-03-29T12:18:08 <dennda> err sorry, wrong channel
2008-03-29T12:18:21 <dennda> good morning, anyway :)
2008-03-29T12:18:27 <johill> dreimark: but I think we need use cases for that first :)
2008-03-29T12:18:29 <johill> dennda: morning :)
2008-03-29T12:18:38 <johill> (in the afternoon. hah)
2008-03-29T12:18:47 <dreimark> johill: got inspired by the refactoring tool. because that can destinguish different definitions in a "text file"
2008-03-29T12:19:18 <johill> huh?
2008-03-29T12:19:20 <dreimark> and you can easily rename only methods
2008-03-29T12:20:14 <johill> ah ok
2008-03-29T12:20:59 <johill> well, like I said, it's not really easy to have links follow a page rename even if you have unique IDs because links aren't written into the page text as unique IDs
2008-03-29T12:21:57 <dreimark> but we can have for the page metadata collecting the links with name and uniq id
2008-03-29T12:22:00 <johill> and doing a id->pagename, pagename->id transformation on every change seems rather prone to breakage
2008-03-29T12:22:17 <johill> even that doesn't really work since you can link to pages that don't exist (don't have an ID) yet
2008-03-29T12:22:51 <johill> plus it would be rather strange
2008-03-29T12:22:55 <johill> say you have a page A that links to B
2008-03-29T12:22:58 <johill> now somebody renames B to C
2008-03-29T12:23:06 <johill> and creates a new B
2008-03-29T12:23:14 <johill> now A says [[B]] but if you click it you go to B
2008-03-29T12:23:16 <johill> eh, C
2008-03-29T12:25:30 <dreimark> hmm, yes, while refactoring you don't do that. So may be we can do it only offline by a moin subcommand.
2008-03-29T12:26:20 <johill> that could then go through all pages and update the page text
2008-03-29T12:26:31 <johill> so for that you don't even need a unique ID
2008-03-29T12:26:51 <johill> anyway, lunch
2008-03-29T12:27:19 <dreimark> oh yeah me too, forgot totally the time
2008-03-29T13:17:07 <HongJun> hello
2008-03-29T13:18:44 <HongJun> ThomasWaldmann: Hi, are you still online?
2008-03-29T13:22:21 <ThomasWaldmann> yes
2008-03-29T13:25:00 <HongJun> ThomasWaldmann: when i write my application for the "MoinMoin RESTful Interface", I find a problem that up to now, most browsers don't support PUT and DELETE in Web forms
2008-03-29T13:27:14 <HongJun> ThomasWaldmann: If the moin wil face to provide web servic, the RESTful Interface could be useful, I am not sure whether moin will go that way
2008-03-29T13:38:08 <ThomasWaldmann> well, as I said, you will have to show in your application how that is useful and how it would look like - in a very concrete way
2008-03-29T13:44:08 <ThomasWaldmann> the more changes to existing apis that change would require and the less practical use / practical advantages it gives, the less likely we would do such a change
2008-03-29T13:44:26 <HongJun> ThomasWaldmann: Is moin going to provide web servic onday
2008-03-29T13:45:08 <ThomasWaldmann> maybe, if it is useful? see above :)
2008-03-29T13:45:36 <HongJun> ThomasWaldmann: can I still apply it as SoC
2008-03-29T13:46:09 <ThomasWaldmann> you are completely free what you put into your application(s)
2008-03-29T13:48:40 <HongJun> ThomasWaldmann: do you mean that I can't apply it for SoC? Maybe I should change to anothor project
2008-03-29T13:51:19 <ThomasWaldmann> maybe update your application with regards to what I have told you
2008-03-29T13:51:43 <ThomasWaldmann> then ask for comments again (me or other people here)
2008-03-29T13:52:07 <ThomasWaldmann> if you can't show how it would be useful, it maybe is not
2008-03-29T13:52:33 <HongJun> ok, thank you
2008-03-29T13:52:46 <dreimark> HongJun: we want to see a plan and that in a concrete way
2008-03-29T13:53:07 <dreimark> to get how much changes and what's the practical usage
2008-03-29T13:53:26 <dreimark> ThomasWaldmann has told that at all
2008-03-29T13:55:11 <dreimark> another application added
2008-03-29T13:56:13 <dreimark> but content is missing
2008-03-29T13:57:07 * waldi produces locking problems ...
2008-03-29T13:57:25 <johill> waldi: ?
2008-03-29T13:59:00 <waldi> johill: i thought about the page rename problem and wanted to propose something like: page in the storage gets links with name and id. the id is stripped during edit and readded during save. but it presumes that the page got not renamed during the edit
2008-03-29T13:59:32 <johill> ah, about dreimark's thing
2008-03-29T13:59:45 <waldi> hmm, does current browsers support svg embedded into an xhtml file?
2008-03-29T13:59:48 <xorAxAx> johill: i completed reading the file, makes sense to me
2008-03-29T14:00:11 <johill> xorAxAx: :)
2008-03-29T14:01:01 <johill> waldi: I don't think so
2008-03-29T14:01:27 <waldi> mathml is supported, at least by gecko
2008-03-29T14:01:43 <johill> yeah, but svg you have to link with <img> tags iirc
2008-03-29T14:01:48 <waldi> no
2008-03-29T14:01:57 <waldi> it is xml
2008-03-29T14:02:21 <johill> yeah, still. might be better to just try :)
2008-03-29T14:02:27 <waldi> according to http://wiki.svg.org/Inline_SVG it seems to work
2008-03-29T14:02:38 <waldi> for browsers, not for browser emulations
2008-03-29T14:03:26 <johill> heh
2008-03-29T14:03:47 <xorAxAx> well, xhtml doesnt allow a lot of namespaces
2008-03-29T14:03:49 <johill> xorAxAx: best just add any comments to the file, I hvae to go in about 20 minutes
2008-03-29T14:03:52 <xorAxAx> one of the allowed ones is svg, though
2008-03-29T14:03:53 <waldi> xorAxAx: xml does
2008-03-29T14:04:08 <waldi> you can push any namespace in it
2008-03-29T14:04:13 <xorAxAx> waldi: the point was that xhtml supports it natively
2008-03-29T14:05:21 <waldi> supports what?
2008-03-29T14:05:30 <xorAxAx> waldi: svg in the document
2008-03-29T14:05:35 <xorAxAx> btw, it looks like a deadline extension is likely
2008-03-29T14:05:41 <johill> heh
2008-03-29T14:05:56 <johill> xorAxAx: you mentioned that my stuff wasn't concrete enough, what should I add?
2008-03-29T14:05:59 <waldi> xorAxAx: no, my point was, if the useragents supports it
2008-03-29T14:06:21 <xorAxAx> waldi: i know, thats the reality level. of course its completly useless in reality
2008-03-29T14:06:46 <xorAxAx> johill: hmm
2008-03-29T14:06:56 <waldi> well, anything except IE supports it. IE does not even support svg at all
2008-03-29T14:07:14 <xorAxAx> johill: i guess its fine from my point of view now, maybe ask dennda if he needs some more input in case he would have been chosen to work on this :-)
2008-03-29T14:07:24 <xorAxAx> waldi: there is an adobe plugin
2008-03-29T14:07:33 <waldi> xorAxAx: sure
2008-03-29T14:07:45 <johill> xorAxAx: ok. I guess I could also elaborate a bit on teh mutable data part
2008-03-29T14:07:57 <johill> xorAxAx: and we need to come to an agreement about edit-locks :)
2008-03-29T14:08:00 <xorAxAx> johill: and fix the sections (e.g. the acl one) :)
2008-03-29T14:08:08 <johill> fix?
2008-03-29T14:08:09 <xorAxAx> johill: well, do you disagree with my idea? :)
2008-03-29T14:08:20 <xorAxAx> well, it contained irrelevant thoughts about legacy code IMHO
2008-03-29T14:08:45 <johill> ah that, actually, upon re-reading it it is fine as long as you assume it's talking about 'current storage' as the legacy code ;)
2008-03-29T14:09:27 <johill> xorAxAx: I'm not really sure yet, how would you detect that two people tried to create a new page at the same time and warn the second?
2008-03-29T14:09:41 <xorAxAx> johill: you dont warn the second one before saving
2008-03-29T14:09:47 <xorAxAx> johill: but merge fine
2008-03-29T14:10:02 <johill> how do you merge two completely new pages?
2008-03-29T14:10:30 <johill> and doesn't that conflict with adminstrator expectations when you have hard edit locks in place?
2008-03-29T14:10:47 <xorAxAx> johill: well, like diff3
2008-03-29T14:11:09 <xorAxAx> johill: yes, a bit. but thats a kind of issue i would be willing to take :)
2008-03-29T14:11:23 <xorAxAx> given that i remember not a single bug report about that :)
2008-03-29T14:11:34 <xorAxAx> (the fact that a warning is missing, not that there is some data loss)
2008-03-29T14:12:00 <xorAxAx> i can only imagine a pretty few cases where this would happen IRL
2008-03-29T14:12:52 <dreimark> xorAxAx: and what happens if the first one saved before with some acls protecting the save of the second one?
2008-03-29T14:13:53 <xorAxAx> dreimark: very good edge case ... i guess it would need to be handled in the merging logic because it can happen in any other context as well
2008-03-29T14:14:06 <xorAxAx> (i.e. even if the page is not new
2008-03-29T14:15:22 <johill> xorAxAx: true
2008-03-29T14:15:47 <johill> xorAxAx: diff3 is pretty useless if you start from an empty file
2008-03-29T14:15:58 <johill> (unless you happen to write really similar things)
2008-03-29T14:15:59 <xorAxAx> johill: no, it produces a consistent and expected result
2008-03-29T14:16:03 <waldi> johill: it will just append both contents
2008-03-29T14:16:21 <johill> yeah which is pretty useless if you describe the same thing with different words
2008-03-29T14:16:33 <xorAxAx> well, this is not a decidable problem :-)
2008-03-29T14:16:35 <johill> anyway, we can also just detect that and pretend it confliceted
2008-03-29T14:16:41 <johill> and ask the user
2008-03-29T14:17:18 <xorAxAx> ?
2008-03-29T14:17:24 <xorAxAx> johill: thats the idea
2008-03-29T14:17:33 <xorAxAx> you must have misunderstood me
2008-03-29T14:17:45 <xorAxAx> the idea is to resolve every conflict the same
2008-03-29T14:18:10 <johill> oh I was under the impression that if diff3 didn't give you a conflict it would just be saved
2008-03-29T14:18:51 <CIA-39> Johannes Berg <johannes AT sipsolutions DOT net> default * 3771:cb1163df537c 1.7-storage-hwendel/LOCKFREE: remove some acl checking 'update' thoughts, just confusing
2008-03-29T14:19:24 <johill> so in that case, yes, we just disallow having an item w/o revisions and don't use item metadata for it ever
2008-03-29T14:19:57 <johill> that solves this particular problem, we'll still need a lock for item metadata updates (even a read lock for it to avoid the consistency problem) but that should be fine
2008-03-29T14:20:05 <johill> we'll need similar locks for anything that is mutable
2008-03-29T14:20:17 <xorAxAx> a read lock?
2008-03-29T14:20:39 <xorAxAx> well, not if you push updates to the item metadata atomicly
2008-03-29T14:20:40 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3426:79dd730d6455 1.7/wiki/htdocs/ (3 files in 3 dirs): common.css: added yellow and orange as background color
2008-03-29T14:20:48 <xorAxAx> like its done currently (everything accepts dicts)
2008-03-29T14:21:02 <johill> hm, true
2008-03-29T14:21:10 <johill> why did I think you need a read lock then?
2008-03-29T14:21:23 <xorAxAx> when did you think that?
2008-03-29T14:21:27 <johill> not sure :)
2008-03-29T14:21:37 <xorAxAx> ah, wait
2008-03-29T14:21:40 <johill> I think I confused things a bit, you need to take the write lock before reading the data to change
2008-03-29T14:21:47 <xorAxAx> yes
2008-03-29T14:22:11 <xorAxAx> dirty read is not an issue because of the bulk update
2008-03-29T14:22:25 <johill> but that's checked even in the current code with the ReadonlyMetadata object
2008-03-29T14:23:33 <johill> yeah, I'm pretty sure now I confused that with the read part of an update
2008-03-29T14:24:11 <johill> so yes, as long as we don't need to remove_item because somebody canceled an edit, which implies not saving anything for an edit of a new page, which implies no edit-locking in its metadata, we're fine
2008-03-29T14:24:27 <johill> we could still do that, in fact, if we simply allowed having empty items be persistent
2008-03-29T14:24:39 <johill> i.e. simply leave the empty directory on the filesystem there
2008-03-29T14:25:06 <johill> but that makes exists() a higher-level check because the backend will say "sure, that thing is there" while the frontend really cares "does it have any revisions"
2008-03-29T14:25:21 <xorAxAx> """Applications to PHP and Gentoo haven't been very encouraging so far -
2008-03-29T14:25:21 <xorAxAx> only 10 each. Quality-wise, we have got only 1 or 2 really good ones,
2008-03-29T14:25:22 <xorAxAx> so +1 for an extension."""
2008-03-29T14:25:50 <johill> heh
2008-03-29T14:26:01 <johill> anyway, I have to go, I'll be off for the weekend.
2008-03-29T14:26:06 <xorAxAx> c u
2008-03-29T14:26:14 <johill> cu
2008-03-29T14:27:07 <xorAxAx> btw, it looks like the mails from the google app only go to the org admins in case of no mentor being assigned yet
2008-03-29T14:30:30 <HongJun> ThomasWaldmann: I hava made some change on my wiki, please review it
2008-03-29T14:33:46 <xorAxAx> HongJun: do you want to apply for gsoc?
2008-03-29T14:34:31 <HongJun> xorAxAx: yep
2008-03-29T14:35:25 <xorAxAx> HongJun: then you might want to see your text reviewed by more than person :)
2008-03-29T14:36:45 <HongJun> xorAxAx: I am not very clear about what you mean
2008-03-29T14:38:02 <xorAxAx> HongJun: you might want to paste an URL here and be open to questions from other mentors as well
2008-03-29T14:41:04 <HongJun> http://moinmo.in/HongJun
2008-03-29T15:05:46 <dreimark> bbl
2008-03-29T16:24:22 <zenhase> btw, what is florian festi doing right now? is he still somehow involved?
2008-03-29T16:28:48 <xorAxAx> zenhase: as much involved as one year ago
2008-03-29T16:28:56 <xorAxAx> --> not at all :)
2008-03-29T16:31:17 <zenhase> well, he visited from time to time ;)
2008-03-29T16:31:34 <xorAxAx> zenhase: here?
2008-03-29T16:32:07 <zenhase> nevermind :P
2008-03-29T16:34:53 <xorAxAx> zenhase: well, i know they you met him
2008-03-29T16:35:50 <xorAxAx> HongJun: so, you describe the REST idea on that page. what would be your deliverables?
2008-03-29T16:38:27 <xorAxAx> hehe, a fourth application
2008-03-29T16:40:04 <zenhase> fourth already?
2008-03-29T16:40:10 <zenhase> oh :)
2008-03-29T16:40:28 <zenhase> and i am working on my two applications
2008-03-29T16:41:55 <xorAxAx> its not very useful yet
2008-03-29T16:42:04 <xorAxAx> lets see if he will extend it
2008-03-29T16:42:43 <HongJun> xorAxAx: an interface for moin to rest
2008-03-29T16:43:11 <xorAxAx> HongJun: please elaborate on the wiki page, what the components of such an interface are
2008-03-29T16:46:43 <HongJun> xorAxAx: I am not very clear about the components right now, it needs time to make it out, maybe a set of API
2008-03-29T16:47:42 <HongJun> xorAxAx: Provide the simple API for other application to use moin data
2008-03-29T16:48:42 <xorAxAx> HongJun: well, your dispatcher calls methods. where do these methods come from? will you implement them? if yes, how would they look like? please elaborate a bit on that
2008-03-29T16:51:33 <HongJun> xorAxAx: dispatch requests to CRUD methods.
2008-03-29T16:56:36 <HongJun> xorAxAx: Maybe it need to redesign the Action model
2008-03-29T16:59:05 <HongJun> xorAxAx: By defining a base class "Resouce", and then let all the resources derived from it
2008-03-29T17:11:38 * ThomasWaldmann will create master{16,17}.moinmo.in now except there are better suggestions
2008-03-29T17:16:33 <xorAxAx> ThomasWaldmann: well, how do you like the ones that i made 3 years ago?
2008-03-29T17:16:40 <xorAxAx> IMHO they are more feasible than forking
2008-03-29T17:16:51 <xorAxAx> let me search the wikipage
2008-03-29T17:17:02 <xorAxAx> http://moinmo.in/MoinMasterBranches
2008-03-29T17:19:33 <ThomasWaldmann> what exactly do you suggest?
2008-03-29T17:20:14 <xorAxAx> some mixture of the 2nd/3rd solution probably
2008-03-29T17:21:08 <ThomasWaldmann> well, no code exists for either one
2008-03-29T17:21:50 <xorAxAx> writing it is a small task if the decision was done
2008-03-29T17:21:50 <ThomasWaldmann> (not even speaking of practically tested code and procedures)
2008-03-29T17:22:25 <xorAxAx> well, for the last 3 years you didnt signal any interest in these solutions
2008-03-29T17:23:26 <ThomasWaldmann> well, I am mostly interested in proven working stuff :)
2008-03-29T17:23:52 <ThomasWaldmann> and that approach won't work for MoinI18n/*
2008-03-29T17:23:55 <xorAxAx> aha, never had that impression :)
2008-03-29T17:24:05 <xorAxAx> yes, good point
2008-03-29T17:24:19 <xorAxAx> well, there could be 2 copies of that subhierachy
2008-03-29T17:26:14 <ThomasWaldmann> same for InstallDocs
2008-03-29T17:26:28 <ThomasWaldmann> -> won't work
2008-03-29T17:27:08 <ThomasWaldmann> any other suggestions?
2008-03-29T17:27:18 <xorAxAx> i dont see the problem with install docs
2008-03-29T17:27:26 <xorAxAx> please elaborate
2008-03-29T17:30:30 <xorAxAx> ThomasWaldmann:
2008-03-29T17:32:24 <ThomasWaldmann> if we want to change contents more than adding a few words/lines, we would end up putting numbers into pagenames there also
2008-03-29T17:33:03 <xorAxAx> why? the 3rd solution would make it easy to accomodate both versions
2008-03-29T17:33:12 <xorAxAx> also i dont see why there would be large changes in the docs
2008-03-29T17:33:41 <ThomasWaldmann> e.g. if we do the wsgi refactoring
2008-03-29T17:34:11 <xorAxAx> why would that end up in different docs?
2008-03-29T17:34:25 <ThomasWaldmann> why not?
2008-03-29T17:34:26 <xorAxAx> i posed you this question a few times in the last years and still havent got an answer
2008-03-29T17:34:38 <xorAxAx> because you can still keep up the old interfaces for the fcgi, mod_py etc. servers
2008-03-29T17:34:47 <xorAxAx> the same is true for standalone
2008-03-29T17:35:06 <xorAxAx> and the wsgi docs are missing, in 1.6 and 1.7
2008-03-29T17:35:07 <ThomasWaldmann> what's would be the point of it, if we keep all the old cruft?
2008-03-29T17:35:22 <xorAxAx> because somebody merged wsgi without checking for docs
2008-03-29T17:35:30 <xorAxAx> ThomasWaldmann: much nicer code in moin
2008-03-29T17:35:54 <xorAxAx> and you need the old features anyway (you want to support fcgi etc.)
2008-03-29T17:36:00 <xorAxAx> so there is not much cruft to keep
2008-03-29T17:36:21 <ThomasWaldmann> well, I suggest you add more details to the wsgi refactoring page
2008-03-29T17:36:57 <xorAxAx> but thats orthogonal to the moinmaster stuff a bit
2008-03-29T17:37:17 <xorAxAx> so, do you still think that installation docs couldnt be kept with the described scheme?
2008-03-29T17:37:26 <ThomasWaldmann> my idea of wsgi was to drop lots of code and suggest users to use external protocol converting middleware
2008-03-29T17:38:00 <xorAxAx> well, as i said, its important to ship a bit of support code, like trac does
2008-03-29T17:38:01 <ThomasWaldmann> as soon as we want to do bigger changes, we would have a problem
2008-03-29T17:38:46 <ThomasWaldmann> because changes in dev would also change docs generation for stable
2008-03-29T17:39:23 <ThomasWaldmann> xorAxAx: update that page
2008-03-29T17:39:39 <xorAxAx> hmm, can you explain your last sentence?
2008-03-29T17:39:53 <ThomasWaldmann> (except if you want to waste time talking about wsgi ever and ever again)
2008-03-29T17:40:28 <xorAxAx> will have a look at the page tonight
2008-03-29T17:42:06 <ThomasWaldmann> btw, if we need a new system page for dev, that page would also be in underlay of a stable update
2008-03-29T17:43:04 <xorAxAx> well, the feature parser could be used page-wide :)
2008-03-29T17:43:13 <xorAxAx> with 2 new lines of code in the flatten script
2008-03-29T17:43:52 <xorAxAx> but indeed, one would want to diff the page trees before checking them in to see whether somebody added 1.7 related stuff to the 1.6 set
2008-03-29T17:45:37 <ThomasWaldmann> I'll just clone it.
2008-03-29T17:45:48 <xorAxAx> hmm?
2008-03-29T17:45:55 <xorAxAx> you mean fork?
2008-03-29T17:45:58 * xorAxAx sighs
2008-03-29T17:46:19 <xorAxAx> hmm, on the other hand - it depends on how 1.6 will evolve
2008-03-29T17:46:40 <xorAxAx> so if this minimalist strategy will continue to work, maybe its not so bad
2008-03-29T17:48:25 <ThomasWaldmann> maybe realize that there is no other way right now anyway
2008-03-29T17:48:51 <ThomasWaldmann> I won't rely on non-existing code and unproven procedures...
2008-03-29T17:48:55 <xorAxAx> if you feel blocked by the scripts, indeed. but i can finish them in 30 minutes probably. but i guess i dont want to :)
2008-03-29T17:49:04 <xorAxAx> well, forking master isnt unproven either ;-)
2008-03-29T17:49:15 <xorAxAx> but go ahead
2008-03-29T17:50:31 * xorAxAx -> jogging
2008-03-29T17:53:11 <ThomasWaldmann> http://www.lolcat.de/images/36278.jpg
2008-03-29T18:11:30 <dreimark> 5 application now
2008-03-29T18:15:15 <zenhase> ThomasWaldmann: uhm, thanks for the anti-spam correction :)
2008-03-29T18:16:08 <dreimark> johill: http://www.egenix.com/library/moinmoin/
2008-03-29T18:16:26 <dreimark> see Search And Replace Multiple Pages Action Plugin
2008-03-29T18:20:58 <dennda> uff still so few? :(
2008-03-29T18:21:02 <dennda> zenhase: did you already applyÃ
2008-03-29T18:21:24 <dennda> (that was supposed to be a question mark)
2008-03-29T18:22:02 <dreimark> dennda: he is working on it
2008-03-29T18:22:11 <dennda> ah ok
2008-03-29T18:22:24 <dennda> Maybe I'll file another application to improve my chances :)
2008-03-29T18:22:52 <dreimark> good idea
2008-03-29T18:23:42 <dreimark> i guess there is a relation in filed applications anf slots we get
2008-03-29T18:26:36 <dennda> dreimark: yes, xorAxAx told me that
2008-03-29T18:26:59 <dennda> so I encourage everyone to do the same ;)
2008-03-29T18:33:11 <zenhase> dennda: not really
2008-03-29T18:33:20 <zenhase> dennda: i am writing my applications right now
2008-03-29T18:33:29 <zenhase> ah, dreimark said it already :)
2008-03-29T18:38:15 * gizmo is writing her application too :
2008-03-29T18:40:31 * ThomasWaldmann thinks about sharing user_dir between 1.6 and 1.7
2008-03-29T18:47:11 <dreimark> gizmo: :)
2008-03-29T18:50:42 <dennda> zenhase: How many do you want to file?
2008-03-29T18:51:49 * ThomasWaldmann .oO(quality over quantity)
2008-03-29T19:04:48 <dreimark> bbl
2008-03-29T19:05:48 <gizmo> dreimark: :)
2008-03-29T19:13:41 <ThomasWaldmann> ok, master16.moinmo.in and master17.moinmo.in are online
2008-03-29T19:28:06 <zenhase> dennda: i am going to file 2
2008-03-29T19:28:39 <zenhase> ThomasWaldmann: yeah, it's not about the quantity, it's just two different parts that got my attention
2008-03-29T19:29:04 <zenhase> boy, right now i just finished the first abstract o_O
2008-03-29T19:29:15 <zenhase> now i need the detailled outline
2008-03-29T19:29:19 <zenhase> *work work*
2008-03-29T19:29:47 <dennda> zenhase: which topics?
2008-03-29T19:35:23 <zenhase> send_page and WSGI
2008-03-29T19:38:46 <dennda> ok
2008-03-29T19:38:52 <dennda> I think I'll file another one for doodle
2008-03-29T19:43:54 <ThomasWaldmann> http://master16.moinmo.in/MoinI18n/ar <-- look there then navigate to same page in master17 (see coloured box at top) and watch how rendering changes
2008-03-29T19:45:22 <ThomasWaldmann> (does that happen on firefox2 also?)
2008-03-29T19:45:53 <xorAxAx> ThomasWaldmann: can you replace <u> in the box with something more sane? :)
2008-03-29T19:46:03 <xorAxAx> <big> or something like that :)
2008-03-29T19:46:27 <xorAxAx> here, the rendering matches besides the box color (opera)
2008-03-29T19:46:44 <xorAxAx> and you didnt configure mail in the 1.6 wiki
2008-03-29T19:50:09 <ThomasWaldmann> xorAxAx: you mean 1.7?!
2008-03-29T19:50:55 <xorAxAx> ThomasWaldmann: hmm, oops
2008-03-29T19:51:11 <xorAxAx> ThomasWaldmann: now i know what confused me - the session data wasnt valid on the 1.6 wiki
2008-03-29T19:51:18 <xorAxAx> but i am logged in on the 1.7 one
2008-03-29T19:51:28 <xorAxAx> hmm, no
2008-03-29T19:51:39 <xorAxAx> ah, the 1.7 has anonymous sessions, thats it :)
2008-03-29T19:51:42 <xorAxAx> nevermind
2008-03-29T19:51:55 <xorAxAx> hmm, but even then, why is there a subscribe link?
2008-03-29T19:52:02 <xorAxAx> bug
2008-03-29T19:52:09 <xorAxAx> (in 1.7)
2008-03-29T19:54:41 <ThomasWaldmann> both wikis had email enabled, but one with outdated values
2008-03-29T19:55:22 * ThomasWaldmann switches the 1.7 stuff to r/w now
2008-03-29T19:56:38 <ThomasWaldmann> ehrm, already was X)
2008-03-29T20:03:58 <dreimark> xorAxAx: you need to login to use but of course it should not be shown in the editbar
2008-03-29T20:04:09 <xorAxAx> dreimark: yep, thats the bug
2008-03-29T20:04:33 <dreimark> i tried to be sure we have not two bugs
2008-03-29T20:08:06 <xorAxAx> ah
2008-03-29T20:12:50 * dreimark fixes that
2008-03-29T20:12:59 <ThomasWaldmann> dreimark: wait
2008-03-29T20:14:30 <ThomasWaldmann> the code had some stuff to switch off actions in the past. the problem was it switched off too much, so you could do some stuff only by editing the url. don't reintroduce THAT BUG again.
2008-03-29T20:18:42 <dreimark> ThomasWaldmann: won't do in that case it is just the wrong logic
2008-03-29T20:18:46 <dreimark> http://paste.pocoo.org/show/36351/
2008-03-29T20:19:33 <dreimark> mine is not ready too hehe
2008-03-29T20:20:29 <ThomasWaldmann> just add parantheses around the or
2008-03-29T20:25:49 <dreimark> dumb negative tests
2008-03-29T20:36:24 <zenhase> applications += 1
2008-03-29T20:36:28 <zenhase> ;)
2008-03-29T20:36:50 <gizmo> zenhase: ;)
2008-03-29T20:37:02 <zenhase> i take a break eating dinner now
2008-03-29T20:37:11 <zenhase> then i will take on the WSGI application
2008-03-29T20:40:01 <zenhase> oh, and i can put the application on the wiki too then, i think peer review makes fleshing it out more easy
2008-03-29T20:42:50 <xorAxAx> makes sense, we will inform you about students who copy the wiki page into a new application ;-)
2008-03-29T20:43:27 <dennda> good idea!
2008-03-29T20:43:33 <dennda> :)
2008-03-29T20:43:39 * gizmo confused
2008-03-29T20:44:08 <xorAxAx> dennda: psst, dont forget to paraphrase every second sentence :-)
2008-03-29T20:45:38 <gizmo> xorAxAx: me me me :D ;)
2008-03-29T20:46:01 <zenhase> xorAxAx: well, actually this application is quite specific i think :)
2008-03-29T20:46:11 <xorAxAx> gizmo: hmm? :)
2008-03-29T20:46:24 <zenhase> xorAxAx: i don't expect any other students applying for that one
2008-03-29T20:46:55 <gizmo> xorAxAx: my application :D
2008-03-29T20:47:01 <xorAxAx> gizmo: be patient ;-)
2008-03-29T20:47:05 <gizmo> zenhase: for what project
2008-03-29T20:47:08 <zenhase> i mean it's more like a loose end i feel, that's leftover from last year
2008-03-29T20:47:31 <gizmo> xorAxAx: well, ok
2008-03-29T21:39:48 <ThomasWaldmann> johill: http://master17.moinmo.in/4ct10n/info/HelpOnConfiguration?action=diff&rev2=212&rev1=210 why did you change cookie_lifetime description?
2008-03-29T21:44:43 * ThomasWaldmann adds the old desc again
2008-03-29T22:29:25 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3427:0cc5867d57f5 1.7/MoinMoin/theme/_tests/test_editbar_actions.py: theme._test.test_editbar_actions: tests actions for editbar dependend on valid user and email enabled
2008-03-29T22:29:27 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3428:a53f1e77d372 1.7/MoinMoin/theme/__init__.py: theme.__init__: fixed bug for subscribe link
2008-03-29T22:29:28 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3429:9cdac4c5843e 1.7/MoinMoin/macro/_tests/test_Hits.py: macro._tests.test_Hits: fixed test by removing cache
2008-03-29T22:29:29 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3430:c06b749e9c68 1.7/MoinMoin/script/maint/cleancache.py: MoinMoin.script.maint.cleancache: removes cache based on event-log
2008-03-29T22:29:31 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3431:b64bd80bd53c 1.7/MoinMoin/ (3 files in 3 dirs): fixed some doc strings
2008-03-29T22:50:45 <dreimark> TheSheep: I am not sure if I remember right but I think you have converted some of the macros listed on http://moinmo.in/MoinMoinTodo/Release%201.7.0/ListOfMacros
2008-03-29T22:51:41 <TheSheep> dreimark: Ah, sorry, I'll update the list
2008-03-29T23:02:46 <dreimark> do you have pushed them ?
2008-03-29T23:03:42 <TheSheep> dreimark: I have pushed them
2008-03-29T23:04:57 <TheSheep> http://hg.moinmo.in/moin/1.7/rev/4aeb96183e04 and http://hg.moinmo.in/moin/1.7/rev/55b787c6c083
2008-03-29T23:06:13 <dreimark> :)
2008-03-29T23:21:48 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3432:0b56ed4808a6 1.7/MoinMoin/ (action/newpage.py macro/NewPage.py): action.newpage, macro.NewPage: changed to POST
2008-03-29T23:21:49 <CIA-39> Reimar Bauer <rb.proj AT googlemail DOT com> default * 3433:21d9d7529304 1.7/MoinMoin/ (action/newpage.py macro/Navigation.py macro/NewPage.py): fixed docstring
2008-03-29T23:23:30 <dreimark> action newpage equires now POST
2008-03-29T23:46:17 <dreimark> strange sorting of the Application list MoinMoin: MoinMoinChat/Logs/moin-dev/2008-03-29 (last edited 2008-03-28 23:15:02 by IrcLogImporter)
