IRC log of dig on 2007-08-09
Timestamps are in UTC.
- 00:37:10 [kennyluck]
- kennyluck has quit ()
- 00:43:53 [timbl_]
- timbl_ has quit (Read error: 110 (Connection timed out))
- 00:52:44 [lkagal]
- lkagal (n=lkagal1@79.68.171.66.subscriber.vzavenue.net) has joined #dig
- 01:05:45 [cph_mit]
- cph_mit (n=cph@pool-71-184-112-148.bstnma.fios.verizon.net) has joined #dig
- 01:21:36 [presbrey]
- presbrey (n=presbrey@c-24-61-47-67.hsd1.ma.comcast.net) has joined #dig
- 01:26:35 [presbrey_]
- presbrey_ (n=presbrey@c-24-61-47-67.hsd1.ma.comcast.net) has joined #dig
- 01:26:35 [presbrey]
- presbrey has quit (Read error: 104 (Connection reset by peer))
- 01:45:37 [presbrey_]
- presbrey_ is now known as presbrey
- 02:49:52 [yosi_s]
- yosi_s has quit ("Chatzilla 0.9.77 [Firefox 1.5.0.12/2007073111]")
- 03:29:39 [cph_mit]
- cph_mit has quit ("Leaving.")
- 03:42:24 [timbl]
- timbl (n=timbl@m085e36d0.tmodns.net) has joined #dig
- 03:43:12 [timbl]
- DIGlogger, pointer?
- 03:43:12 [timbl]
- See http://dig.csail.mit.edu/irc/dig/2007-08-09#T03-43-12
- 05:09:17 [timbl]
- timbl has quit (Read error: 110 (Connection timed out))
- 05:46:18 [presbrey]
- presbrey has quit (Read error: 110 (Connection timed out))
- 06:32:40 [kennyluck]
- kennyluck (n=kennyluc@125-226-51-192.dynamic.hinet.net) has joined #dig
- 06:39:12 [kennyluck]
- DIGlogger, pointer?
- 06:39:12 [kennyluck]
- See http://dig.csail.mit.edu/irc/dig/2007-08-09#T06-39-12
- 06:50:08 [kennyluck]
- kennyluck has quit ()
- 09:47:18 [kennyluck]
- kennyluck (n=kennyluc@125-226-51-192.dynamic.hinet.net) has joined #dig
- 11:37:19 [RalphS]
- RalphS (n=swick@30-6-188.wireless.csail.mit.edu) has joined #dig
- 13:11:25 [kennyluck]
- kennyluck has left #dig
- 13:31:41 [jambo]
- jambo (i=jambo@30-6-199.wireless.csail.mit.edu) has joined #dig
- 13:44:43 [ericP]
- ericP (n=ericP@homer.w3.org) has joined #dig
- 13:45:08 [ericP]
- joe's not around?
- 13:45:27 [ericP]
- (not sure if he disguises himself under an alias)
- 13:47:19 [ericP]
- ericP has left #dig
- 14:22:04 [lkagal]
- lkagal has quit ()
- 14:39:37 [yosi_s]
- yosi_s (n=syosi@30-7-176.wireless.csail.mit.edu) has joined #dig
- 14:57:24 [RalphS]
- I confirm 404 at http://dig.csail.mit.edu/issues/tabulator
- 14:57:40 [RalphS]
- there are probably notes in the (inaccessible) tabulator archive on how it's setup
- 14:57:59 [RalphS]
- it's likely something we have the power to fix
- 14:58:06 [RalphS]
- David Sheets would know
- 14:58:18 [jambo]
- yeah, i was going to drop him an e-mail
- 14:58:27 [RalphS]
- I suggesting sending a request to help@csail.mit.edu
- 14:58:44 [jambo]
- okay
- 14:59:03 [RalphS]
- not really their problem, but since it's probably caused by their changes they should help us diagnose and fix
- 15:00:28 [lkagal]
- lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
- 15:02:32 [lkagal]
- lkagal has quit (Client Quit)
- 15:03:17 [lkagal]
- lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
- 15:04:15 [lkagal]
- lkagal has quit (Client Quit)
- 15:05:43 [lkagal]
- lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
- 15:08:45 [jambo]
- DIGlogger, pointer
- 15:08:45 [jambo]
- See http://dig.csail.mit.edu/irc/dig/2007-08-09#T15-08-45
- 15:24:06 [kennyluck]
- kennyluck (n=kennyluc@125-226-51-192.dynamic.hinet.net) has joined #dig
- 15:36:46 [jambo]
- kennyluck, do you have a moment
- 15:36:56 [kennyluck]
- yes
- 15:37:35 [jambo]
- the keyboard behavior for editing in the outline is much better now
- 15:37:54 [kennyluck]
- Okay, thank you.
- 15:37:56 [jambo]
- but the option of just starting to type still needs to be removed because of what i said yesterday
- 15:38:14 [kennyluck]
- Yes, sure.
- 15:38:16 [jambo]
- hitting ctrl+t while a node is highlight will cause it to start editing
- 15:38:27 [jambo]
- but then open a new tab
- 15:38:42 [jambo]
- which is no good :)
- 15:39:03 [kennyluck]
- Uh-huh.
- 15:39:12 [jambo]
- but otherwise, the click and wait stuff and hitting enter works good.
- 15:40:12 [jambo]
- oh, one thing i only just noticed
- 15:40:35 [kennyluck]
- ?
- 15:40:36 [jambo]
- hitting enter on symbols that are objects is opening them in a new outline node still
- 15:41:01 [jambo]
- i'm not sure if tim wanted those to be editable
- 15:41:39 [jambo]
- but if he did, then we'd need to get rid of that behavior
- 15:41:57 [jambo]
- that can wait though, I'm sure doing something like that would be relatively easy.
- 15:42:26 [kennyluck]
- what do you mean? I don't get it.
- 15:42:34 [kennyluck]
- those refer to...?
- 15:42:51 [jambo]
- hmm?
- 15:43:07 [jambo]
- probably best with an example
- 15:43:17 [kennyluck]
- ok, sure.
- 15:43:32 [jambo]
- http://dig.csail.mit.edu/2007/wiki/jambobo.rdf
- 15:43:53 [jambo]
- I should be able to edit the creator, right?
- 15:43:57 [jambo]
- or no
- 15:44:19 [kennyluck]
- you mean the predicate?
- 15:44:23 [jambo]
- no, the object.
- 15:44:53 [jambo]
- specifically, though, the node "James Hollenbach" -- not the literal one
- 15:45:40 [kennyluck]
- yes, then? yeah, that should be editable.
- 15:45:46 [jambo]
- okay, well
- 15:45:56 [jambo]
- what combination of keys makes me able to edit it?
- 15:46:30 [kennyluck]
- You mean delete the triple xxx.rdf dc:creator jambo?
- 15:46:59 [kennyluck]
- Or you meant edit the URI directly?
- 15:47:04 [jambo]
- no, I want to change it to say "Ringo Starr" or something
- 15:47:15 [jambo]
- and no longer say jambo
- 15:47:30 [jambo]
- where ringostarr has URI example.com/foo#ringo
- 15:47:34 [jambo]
- that we know about already
- 15:47:46 [kennyluck]
- I see...I would say no and recommend you delete and add a new triple.
- 15:47:59 [jambo]
- that's a little silly, isn't it?
- 15:48:14 [jambo]
- I can edit a literal in place, why can't I edit the symbol in place
- 15:48:31 [kennyluck]
- I don't know.
- 15:48:43 [jambo]
- does your version let you do that?
- 15:48:55 [kennyluck]
- No.
- 15:49:46 [jambo]
- well, would it be easy to make it do that?
- 15:50:09 [jambo]
- having me delete a triple and add a new one seems like something that the UI ought to hide from me, if i just edit it
- 15:50:14 [jambo]
- the same way that it does with literals
- 15:51:04 [kennyluck]
- it is easy. But I am not sure this is good. Anyway, this is not urgent.
- 15:51:15 [jambo]
- What makes it not good?
- 15:51:18 [jambo]
- If the user can delete a triple
- 15:51:21 [jambo]
- and enter a new triple
- 15:51:28 [jambo]
- that goes exactly where the old one is
- 15:51:34 [jambo]
- and it makes the UI more consistent
- 15:51:56 [jambo]
- it's just silly if when you want to edit a literal you can just go ahead and type it, but when you want to edit a symbol you have to first delete it and then add a new one
- 15:52:43 [jambo]
- the code already exists when you have the user typing an object in when they add a new triple -- why can't that code be in the other place too? I'm trying to make the UI consistent instead of confusing
- 15:53:22 [jambo]
- because somebody using it will go "hey I just changed the label to something else. Now i'm going to change the 'last contributor' to myself"
- 15:54:00 [jambo]
- they should be able to do that with the same general key operations as they did when they changed the literal
- 15:54:35 [kennyluck]
- Maybe you're right. Just be sure to agree with Tim.
- 15:57:06 [kennyluck]
- You can do that if you think this is super-urgent. I don't really mind if anyone edits my code.
- 15:57:36 [jambo]
- what are you currently working on?
- 15:57:49 [jambo]
- the most important thing, we agreed on yesterday, was making the UI cosistent.
- 15:57:53 [jambo]
- *consistent
- 15:57:56 [kennyluck]
- I am doing error callbacks. And there are tons of bugs.
- 15:58:14 [kennyluck]
- menu alignment. drag and drop.
- 15:58:27 [jambo]
- please scratch drag and drop
- 15:58:56 [jambo]
- making the simple UI that does keyboard editing spot on is much more important at this stage than drag and drop
- 16:01:04 [kennyluck]
- I don't really want to get rid of drag and drop. Anyway, that's not on top of my todo list.
- 16:02:19 [jambo]
- if you want drag and drop to be in the extension, you can do it after this release. I'm telling you that the goal for this release is to get the text editing spot on
- 16:03:32 [jambo]
- drag and drop does not work cross-OS currently. without you here, it is very difficult to fully discuss what its behavior should be.
- 16:04:10 [kennyluck]
- Okay.
- 16:04:52 [jambo]
- you should be devoting all of your time to getting an acceptable version of keyboard-based editing into the outline, which includes a consistent UI, which we originally outlined when we met with mc in june.
- 16:05:59 [kennyluck]
- I don't really remember what were discussed at that time.
- 16:06:02 [presbrey]
- presbrey (n=presbrey@c-24-61-47-67.hsd1.ma.comcast.net) has joined #dig
- 16:06:19 [kennyluck]
- Hi, Joe.
- 16:06:27 [presbrey]
- hi kenny
- 16:06:56 [kennyluck]
- Uh...http://dig.csail.mit.edu/2007/wiki/test.rdf#test
- 16:06:58 [kennyluck]
- http://dig.csail.mit.edu/2007/wiki/test.rdf#test
- 16:07:13 [jambo]
- the most important part of the entire meeting was that the UI needs to be consistent with what the user will expect from past experiences editing things.
- 16:07:37 [jambo]
- It was taken for granted that that also meant keeping input from node to node very consistent, in the way I just described to you.
- 16:07:44 [jambo]
- All I need from you is:
- 16:08:08 [jambo]
- 1) the ability to edit literals the way we discussed yesterday
- 16:08:13 [kennyluck]
- Joe, in that example, when you add a triple to bar. foo alters as well.
- 16:08:30 [jambo]
- 2) the ability to edit symbols the way we discussed today
- 16:08:56 [jambo]
- and 3) the ability to add completely new triples the way tim explained yesterday, keeping a rigid process
- 16:09:20 [lkagal]
- lkagal has quit ()
- 16:10:21 [kennyluck]
- what is 3)? not achieved in current version?
- 16:11:23 [ericP]
- ericP (n=ericP@homer.w3.org) has joined #dig
- 16:11:44 [jambo]
- timbl wanted the node labeled "TBD" to never really possibly appear to the user -- that is, if the user abandoned adding the new triple, no node labeled "TBD" would appear -- it would just go away
- 16:12:07 [presbrey]
- kenny, the only thing that needs to be changed for that bug is _statement_context; please don't use labeler or rewrite any other functions
- 16:12:14 [davidli]
- davidli (i=dli@31-35-67.wireless.csail.mit.edu) has joined #dig
- 16:12:16 [lkagal]
- lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
- 16:12:59 [ericP]
- presbrey, i looked at HTTP::Daemon. prob is that it isn't multi-threaded and it blocks on a read call without listening for new connections
- 16:13:10 [kennyluck]
- labeler is just a way to generate context.
- 16:13:39 [kennyluck]
- In my opinion, FP IFP > labeler (locally inverse functional property) > iterate over other triples.
- 16:13:44 [ericP]
- should be no problem if you use it from a library that doesn't simultaneously keep pipelines open while making new socket connections
- 16:13:57 [jambo]
- labeler cannot be used in the sparqlUpdate library, kenny. timbl made this very clear yesterday.
- 16:14:42 [kennyluck]
- Anyway, I just wish Joe and timbl can agree on the algorithm and implement it as soon as possible.
- 16:15:11 [davidli]
- DIGlogger, pointer?
- 16:15:11 [davidli]
- See http://dig.csail.mit.edu/irc/dig/2007-08-09#T16-15-11
- 16:15:17 [presbrey]
- we do agree, Kenny. no names, no labeler!
- 16:15:20 [jambo]
- the agreed already that the labeler will not be used in any context in sparql update.
- 16:15:33 [presbrey]
- thanks ericP.
- 16:16:14 [kennyluck]
- I'll leave you with that, Joe.
- 16:16:32 [ericP]
- presbrey, hmm, seems like i should be back it down to HTTP/1.0 to make it close the connections anyways
- 16:18:04 [presbrey]
- ok, that might work. the only thing I usually miss in 1.1 is virtual hosting and we don't need that on the backend
- 16:19:10 [ericP]
- i wonder if the 1.0 switch disables that
- 16:19:34 [ericP]
- the client is still free to send the Host: header, and the server is free to respect it
- 16:20:13 [ericP]
- (though, as you point out, we won't notice the difference without writing explicit code)
- 16:24:35 [presbrey]
- got the issue tracker back up
- 16:25:42 [jambo]
- nice
- 16:25:43 [jambo]
- thanks
- 16:27:25 [kennyluck]
- Joe, in case you hadn't noticed, I altered sparqlUpdate.js a little bit yesterday.
- 16:28:10 [kennyluck]
- in insert_statement
- 16:30:10 [presbrey]
- why did you change update_statement back
- 16:30:30 [presbrey]
- can't you use delete_statement?
- 16:31:33 [kennyluck]
- I didn't do anything about update_statement.
- 16:31:44 [kennyluck]
- ?
- 16:32:28 [ericP]
- presbrey, updated algae -- seems to work
- 16:33:00 [presbrey]
- downgrading http? alright, nice fix =)
- 16:33:11 [ericP]
- just the pipelining
- 16:33:30 [ericP]
- otherwise i have to fix their silly code to use select more wisely
- 16:33:39 [presbrey]
- my mistake, kenny. I diff'd against the wrong version
- 16:34:03 [presbrey]
- yeah I think its better though
- 16:34:22 [jambo]
- presbrey, the tracker is freaking out at me again :(
- 16:35:02 [ericP]
- interface should not be sub { $c = $d->accept(); $r = $c->get_request); ... } but instead { $c = $d->get_request); ... } so they can multiplex between accept and read
- 16:35:53 [presbrey]
- Jim, should be fixed again
- 16:36:03 [ericP]
- presbrey, do you use the response from INSERT or DELETE?
- 16:36:19 [jambo]
- okay
- 16:36:24 [presbrey]
- ericP, have it doing dump-default on insert/delete
- 16:36:49 [ericP]
- right now they appear to say "TRUE" all the time, regardless of whether they matched any or all of the existing data
- 16:37:42 [ericP]
- perhaps DELETE should return the number of triples actually deleted (= number found that matched the delete pattern)
- 16:37:58 [presbrey]
- ya might be nice
- 16:38:09 [presbrey]
- I want to add a query argument to reset the store as well, unless its already there?
- 16:38:55 [presbrey]
- and accordingly, a query argument to load a new file
- 16:39:26 [presbrey]
- (wow, and one to cause --dump-default)
- 16:39:50 [ericP]
- WHERE { ?s ?p ?o } DELETE { ?s ?p ?o }
- 16:39:51 [kennyluck]
- Joe and ericP, what is this message? http://dig.csail.mit.edu/2007/wiki/User:Kennyluck#
- 16:40:27 [presbrey]
- I think then I can get away with setting up a pool of 5-10 algae worker children on dig and RRing all the sparql queries for the whole wiki to them
- 16:41:03 [ericP]
- kennyluck, it's impossible to write <http://dig.csail.mit.edu/2007/wiki/Property:Assignment#> as a prefixed name/local name pair as it doesn't end in any valid localname character
- 16:41:24 [ericP]
- presbrey, do you want them to share data?
- 16:41:27 [ericP]
- 'cause that's work
- 16:41:50 [presbrey]
- no; I'll be writing them to disk and version-controlling them
- 16:43:57 [presbrey]
- still only one algae can work an endpoint at a time
- 16:45:17 [presbrey]
- since its sourcing from disk
- 16:45:53 [jambo]
- presbrey, should callbacks be firing properly from sparqlUpdate again? it doesn't look like davidli's are getting called
- 16:46:35 [presbrey]
- checking...
- 16:47:27 [jambo]
- we are trying sparql.set_object(kb.literal(newText, ""), function(u,s,e) { //somecode });
- 16:47:54 [presbrey]
- yeah syntax looks good
- 16:48:28 [jambo]
- so it should be getting called? we get no errors :(
- 16:48:44 [presbrey]
- still checking; just saying what you typed looked good
- 16:49:01 [jambo]
- ok
- 16:50:12 [presbrey]
- yes; callbacks work
- 16:50:36 [jambo]
- okay, back to work :P
- 16:51:05 [presbrey]
- its possible you are saying "this" and not getting what you expec
- 16:51:30 [presbrey]
- callbacks get window scope
- 16:52:25 [jambo]
- nope
- 16:52:28 [jambo]
- we'll keep looking into it
- 16:53:25 [jambo]
- okay, it's getting called -- something on our end
- 16:53:39 [jambo]
- it didn't look like it was at first
- 17:00:10 [presbrey]
- ericP, is algae doing transactions?
- 17:00:50 [presbrey]
- oh nevermind; I'm doing that, I don't need it to too
- 17:09:22 [ericP]
- presbrey, pretty much, though there are holes in it
- 17:09:46 [ericP]
- added a new command line to make dumping the graph more attractive:
- 17:10:01 [ericP]
- --stfu
- 17:10:01 [ericP]
- -u http://mouni.local:8000/sparql/
- 17:10:01 [ericP]
- --dump-default
- 17:10:02 [presbrey]
- yeah I realized since I'm processing sets of commands at a time and saving back to disk, I've already effectively taken care of transactions though inefficiently
- 17:10:22 [presbrey]
- by more attractive do you mean more easily parsed? =)
- 17:11:03 [ericP]
- yeah, stfu politely instructs it to suppress output other than the graph dump
- 17:14:03 [ericP]
- oh right, another piece of the puzzle: the "SHUTDOWN" command
- 17:14:16 [RalphS]
- diglogger, stop
- 17:15:04 [presbrey]
- cool
- 17:15:50 [RalphS]
- except that didn't actually close the logs
- 17:16:03 [ericP]
- presbrey, you in the building?
- 17:16:13 [presbrey]
- no
- 17:16:14 [ericP]
- campus? city?
- 17:16:19 [presbrey]
- central
- 17:16:50 [RalphS]
- diglogger, bye
- 17:16:50 [RalphS]
- I'm logging. I don't understand 'bye', RalphS. Try /msg DIGlogger help
- 17:16:50 [presbrey]
- I can come in... do you want to hack on algae?
- 17:18:12 [RalphS]
- diglogger, sync
- 17:18:20 [ericP]
- presbrey, no need. i think we can do this here
- 17:18:27 [RalphS]
- ah! sync does it :)
- 17:18:47 [ericP]
- i invoked algae as described above, then did some inserts, selects, deletes, selects, and finally a shutdown
- 17:18:56 [ericP]
- at the end, algae dumped the default graph
- 17:19:03 [presbrey]
- ah perfect, lets add a hook for that somewhere often
- 17:19:30 [RalphS]
- I hate to have to force it on a timer.
- 17:19:39 [RalphS]
- just 'cause of AFS
- 17:19:54 [RalphS]
- I should move the process to wherever the current httpd lives
- 17:20:31 [presbrey]
- what does sync do, flush?
- 17:20:39 [RalphS]
- close();open()
- 17:20:54 [RalphS]
- years ago I think I tried flush and that wasn't enough
- 17:21:22 [RalphS]
- somewhere afs isn't updating the volume info
- 17:22:14 [presbrey]
- could you have gotten the default graph in the interim or just at the end?
- 17:23:30 [RalphS]
- diglogger, sync
- 17:23:48 [presbrey]
- I think you should have it open/close every time it writes