18:17:17 [timbl]Hey kenny, you are in Taiwan?
18:17:58 [timbl]If you scoll up in the log you'll see a problem Jim found with the editing.
18:18:09 [timbl]How was your trip?
18:20:22 [kennyluck]Just OK.
18:20:27 [kennyluck]How's everything?
18:21:44 [timbl]good ... jim is still trying to get the release out.
18:22:23 [timbl]There is a problem that it seems if you control/t (say) when something is selected weird things ahppen.
18:22:33 [timbl]See jambo's comments in e loh
18:22:47 [kennyluck]There are three things I should do now. 1. issue around the problem Jim found 2. menu alignment 3.fix drag and drop. Any suggestion on what I should do?
18:23:28 [kennyluck]By the way, any decision on sparqlUpdate.js?
18:23:55 [timbl]Don't make any big changes, just fix bugs.
18:24:06 [timbl]Chat with jambo about the input bug
18:24:12 [timbl]decision?
18:24:15 [kennyluck]I konw, these are all bugs.
18:24:37 [kennyluck]I guess sparqlUpdate.js is not.
18:24:45 [jambo]is not what?
18:25:28 [kennyluck]a bug, it's about architecture.
18:26:35 [timbl]Kenny, what about sparqlUpdate.js?
18:26:43 [timbl]sparqlUpdate.js is a file.
18:27:46 [kennyluck]I just wonder there is issue around modularity.
18:28:18 [kennyluck]I think I should explain a little bit.
18:28:44 [kennyluck]labeler is used there in order to identify a bnode.
18:29:01 [kennyluck]say, WHERE { ?bnode1 foaf:name "Kenny".}
18:29:25 [jambo]kenny, you put that in, didn't you?
18:29:32 [kennyluck]I did.
18:29:39 [jambo]i don't understand why you don't just keep that in your code
18:29:54 [kennyluck]kennyluck: labeler is used there in order to identify a bnode.
18:29:54 [kennyluck][2:29pm] kennyluck: say, WHERE { ?bnode1 foaf:name "Kenny".}
18:30:08 [timbl]Ah, I see. The issue that sparqlUpdate.js references labeler.
18:30:53 [kennyluck]Because it's likely that, in a document, a human readable name does not identify two things.
18:31:32 [timbl]Kenny, 1) it should not be in sparqlUpdate.js as the basic rdf functionality and the tabulator functionality should be at different levels ... the RDF is the AJAR library, in the rdf subdirectory
18:31:43 [timbl]2) Yu can't use a nme as an idntifiers.
18:31:52 [timbl]You can ONLY use a FP or IFP
18:32:02 [kennyluck]Then how do you address a bnode?
18:33:18 [kennyluck]Well, iterate all the triples and look for one that identifies the bnode might be a way...
18:33:18 [timbl_]timbl_ (n=timbl@m255e36d0.tmodns.net) has joined #dig
18:33:41 [kennyluck]But I think using "names" is more efficient.
18:34:06 [jambo]using names isn't using the data
18:34:19 [kennyluck]Why not?
18:34:31 [jambo]because "label" could be something like "Jim"
18:34:51 [jambo]the string Jim does not uniquely identify me at all
18:35:06 [kennyluck] s = This.store.statementsMatching(undefined,undefined,labelTerm,_statement.why);
18:35:06 [kennyluck] if (s.length == 1) context.push(s[0]);
18:35:32 [jambo]we only want to use FP's and IFP's.
18:36:12 [timbl_]Kenny, do not use names.
18:36:34 [presbrey]http://www.w3.org/2002/12/cal/icalSpec is gone; whats the new link?
18:36:41 [timbl_]If you need to identify a bnode, then do use either an incoming FP or outgoing IFP
18:37:10 [kennyluck]What if the user hasn't input a FP or IFP already?
18:37:34 [kennyluck]Say, those geo data?
18:37:37 [timbl_]presbrey, ask danc what happened to http://www.w3.org/2002/12/cal/icalSpec
18:40:02 [timbl_]geo data?
18:40:14 [timbl_]you mean like based_near?
18:40:41 [kennyluck]Uh-huh
18:40:51 [timbl_]I'd be happy to just turn off editing
18:41:09 [timbl_]for things which are not well enough identified.
18:41:31 [timbl_]I don't know whether based_near is functional.
18:41:42 [timbl_]If so, you are OK. I bet it isn't.
18:41:45 [kennyluck]But why not use "names", if s.length==1?
18:41:55 [timbl_]s.length???
18:42:10 [kennyluck]s = This.store.statementsMatching(undefined,undefined,labelTerm,_statement.why);
18:42:10 [kennyluck][2:35pm] kennyluck: if (s.length == 1) context.push(s[0]);
18:42:33 [kennyluck]That is when in a document, a name identifies only a thing.
18:42:40 [timbl_]If the name is unique.
18:43:07 [kennyluck]We should expect that in a short document.
18:43:36 [timbl_]well, it may happen statistically, and it doens't just apply to names.
18:43:59 [timbl_]Names are not particularly more useful than anything else,.
18:44:15 [timbl_]I would use IFPs if you can
18:44:21 [timbl_]Why are you worried about speed here?
18:44:38 [timbl_]Efficiency? Does this just happen when the user edits something?
18:44:49 [kennyluck]Any way, I just think this is not that critical as of now.
18:45:09 [kennyluck]Before there's decision on what algorithm we use.
18:45:46 [timbl_]Well, the algorithm must use an IFP or FP if there is one.
18:46:06 [timbl_]but it is OK to just refuse to edit things with unidentifiable bnodes just now.
18:46:16 [timbl_]We should generate URIs for things.
18:46:37 [kennyluck]These are new features, not bugs...
18:47:42 [kennyluck]1. issue around the problem Jim found 2. menu alignment 3.fix drag and drop. Any suggestion on what I should do first?
18:49:22 [jambo]1. It is very important that the keys, mouse, and input box all behave as they would be expected to in normal UI paradigms.
18:49:56 [jambo]timbl said that he thought that for a while, pressing enter would begin editing a node.
18:50:06 [jambo]but it has since reverted
18:50:42 [kennyluck]Not really...It never happened like that.
18:51:08 [kennyluck]Uh...except Literals
18:51:23 [jambo]Okay, well it still needs to then. I didn't think that it ever did, but if you look through the log, we definitely agreed that that was what was decided when we met with mc very early on.
18:52:00 [jambo]like i said, it is very important that your interface behaves in a way that people are already familiar with
18:53:06 [kennyluck]I see.
18:53:16 [kennyluck]I'll do 1.
18:54:03 [jambo]okay, so an undesirable solution to 1 would be to catch when the user is pressing control -- that is not what we need
18:54:35 [jambo]what we need is the interface that will allow the user to press enter on or click on an already selected node and begin editing it
18:54:51 [jambo]yes, that does mean removing the behavior that a node gets opened in a new display when the user hits enter on it.
18:55:53 [presbrey]Kenny, are the two fire functions (1 is yours) in sparqlUpdate different in some way or did you copy-paste?
18:56:45 [kennyluck]I copy-paste, sorry.
18:57:05 [timbl_]Jambo, I now know get Error: tabulator.ns.link is not a function
18:57:06 [timbl_]Source File: chrome://tabulator/content/js/rdf/sources.js
18:57:06 [timbl_]Line: 394
18:57:36 [timbl]timbl has quit (Read error: 110 (Connection timed out))
18:57:55 [lkagal]lkagal has quit ()
18:57:58 [timbl_]which means that when sf i guess is called from within tabulator, for some reason the tabulator.ns is not yet set up
18:58:53 [jambo]correct.
18:58:59 [jambo]okay
18:59:03 [jambo]in xpcom.js
18:59:05 [kennyluck]By the way, just out of curiosity, Joe, are you gonna do GET and POST content negotiation?
18:59:26 [presbrey]on the wiki
18:59:39 [kennyluck]Yeah, is that done?
18:59:41 [timbl_]But I have moved the ns initialization before the call to creat e sf
19:00:31 [jambo]okay, hold on timbl
19:00:36 [lkagal]lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
19:01:14 [jambo]okay, timbl, you put it above line 55 in xpcom.js i guess
19:02:05 [timbl_]I have moved it north
19:02:05 [jambo]and that didnt work, yes?
19:02:12 [jambo]okay
19:02:12 [timbl_]Sending xpcom.js
19:02:14 [timbl_]Transmitting file data .
19:02:14 [timbl_]Committed revision 3715.
19:03:12 [timbl_]The error changed from something else to 'is not a function'
19:03:31 [timbl_]can xpcom refer to the rdf stiff?
19:03:43 [jambo]timbl,
19:03:48 [jambo]err hold on
19:04:08 [lkagal]lkagal has quit (Client Quit)
19:05:00 [timbl_]I guess we may need to make a local Namespace function
19:05:10 [jambo]no,
19:05:25 [jambo]that shouldn't be necessary. wait one moment
19:05:29 [presbrey]what exactly do you want it to do, Kenny?
19:07:01 [jambo]timbl, for some reason im not getting the error Namespace is not defined, but I really should be.
19:07:04 [jambo]hold on..
19:07:51 [timbl_]Maybe function Namespace2(nsuri) {
19:07:52 [timbl_] return function(ln) { return tabulator.kb.sym(nsuri+ln) }
19:07:52 [timbl_]}
19:08:03 [timbl_]would be better locally in xpcom
19:08:22 [jambo]It shouldn't be necessary
19:08:25 [jambo]should it?
19:08:39 [jambo]anyway, if you define that there
19:08:56 [jambo]its probably skipping each of those automatically for me because
19:08:59 [jambo]kb is not defined.
19:09:12 [timbl_]Ha
19:09:14 [jambo]the unfortunate part about debugging code in that area
19:09:25 [jambo]is that it will _not_ show up in the error console no matter what
19:09:35 [jambo]try inserting your tabulator.ns stuff
19:09:42 [jambo]between the definition of kb and sf.
19:09:52 [timbl_]Ok
19:10:53 [timbl_]Improvement!
19:11:39 [jambo]what now
19:11:42 [timbl_]Now I get problems in sources-ext.js
19:11:55 [timbl_]Oh .. is the whole sources module forked for the extension??
19:12:02 [jambo]okay, thats a little more marginally to be expected.
19:12:16 [jambo]it has to have a lot of XUL in it
19:12:49 [timbl_]I think the critical line is
19:12:52 [timbl_]tabulator.kb.the(req, tabulator.kb.sym('tab','status'))
19:13:20 [jambo]lstat has no properties huh
19:14:03 [timbl_]means the kb.the failed,
19:14:10 [jambo]yeah
19:14:18 [timbl_]which I assume means the kb.sym gave something silly
19:14:23 [timbl_](though what could it give?)
19:14:32 [jambo]hah
19:14:33 [jambo]okay
19:14:38 [jambo]so I put in a hack yesterday
19:14:42 [timbl_]I have a hacked version of kb.sym which returns tabulator.ns[a](b)
19:14:50 [jambo]okay, did you remove my code
19:14:57 [jambo]that was preventing tab from going through that
19:15:02 [timbl_]I didn't notice your code
19:15:10 [timbl_]I did an update this morning
19:15:13 [jambo]hold on
19:16:53 [jambo]identity.js#132
19:17:18 [jambo]it was because the tabulator ontology was breaking when this was introduced, so i just plugged that in so we could continue testing with it
19:17:36 [kennyluck]Sorry, Joe, I went to the bathroom.
19:18:51 [kennyluck]What I want is to create wiki/User:Kenny#this, so you all can write down my faults there but not here.
19:19:29 [timbl_]jambo, I see line 132. In fact this could happen with any file we load using any prefix which we use, like tab, tabont, owl, rdf, rdds, ect for things we don't use them for
19:19:46 [jambo]yes, i know
19:19:54 [timbl_]My hack was:
19:19:57 [timbl_]RDFFormula.prototype.sym = function(uri,name) {
19:19:57 [timbl_] if (name != null) {
19:19:58 [timbl_] if (!tabulator.ns[uri]) throw 'The prefix "'+uri+'" is not set in the API';
19:19:58 [timbl_]uri = tabulator.ns[uri] + name
19:19:58 [timbl_] }
19:19:58 [timbl_] return new RDFSymbol(uri)
19:19:59 [timbl_]}
19:20:04 [timbl_]Not yet checked in
19:20:15 [jambo]i noted that in the issue too, but i put in that specific one because it let us continue testing
19:20:29 [presbrey]you just want an namespace operator? ":" in addition to "/" ?
19:20:38 [timbl_]I would like to check it in but I can't get it working in the extension
19:20:58 [jambo](it was preventing you from loading resources after the tabulator project was loaded)
19:21:15 [timbl_]The point is that the prefixes in the KB are really not to be used like that
19:21:23 [jambo]yes, i know
19:21:29 [timbl_]the whole use of kb.sym with 2 args is broken
19:21:52 [timbl_]But it is ok if it goes througha n application-wide table which we all agree on
19:22:15 [jambo]okay, well, did removing the catch that i pointed you to change that problem in the extension sources
19:22:17 [timbl_]The hack in my term.js to call out to tabuator.ns is a hack as it breaks the dependency layering
19:23:22 [timbl_]I'll just fix the sources-ext.js to use tabulator.ns.link(...)
19:24:11 [jambo]okay, thats fine
19:24:47 [kennyluck]jambo, timbl, do you have time to discuss some UI issue?
19:24:59 [jambo]sure, what is up
19:25:08 [timbl_]Ok
19:25:32 [timbl_]jambo, extension is running now
19:25:52 [jambo]okay, play around with it a little :)
19:25:53 [kennyluck]Q1: we a TBD is selected, if you type something, should that go into edit mode?
19:26:07 [jambo]no, just make them all act the same way.
19:26:08 [kennyluck]I thought that this is like a spreadsheet application.
19:26:28 [jambo]kenny: the problem is that you're caught halfway between "link" and "spreadsheet"
19:26:51 [kennyluck]But when you type over a Literal, it goes to the edit mode.
19:26:53 [timbl_]Q1:
19:26:56 [timbl_]no.
19:27:19 [jambo]kennyluck, all of the nodes should act the same: hit enter to begin editing it, or click,wait,click to begin editing it
19:27:20 [timbl_]You should have to hit return to get into edit mode, or click an extra time.
19:28:10 [jambo]it is bad for a node containing a symbol to act differently from a node with a literal, because to the user, they can generally look very similar
19:28:12 [kennyluck]Ok, so should I disable the feature that, when you type over a Literal, it goest to the edit mode?
19:28:40 [jambo]yes. Make it so that the user can always both press enter and click-wait-click
19:28:40 [timbl_]Yes.
19:29:02 [kennyluck]Nice, we reach a conclusion.
19:29:12 [timbl_]There should never be user characters accepted as iunput unless there is a visible enput field which has the focus
19:29:17 [jambo]i'm pretty sure it was reached in june ;)
19:29:43 [timbl_]Yes, mc schraeful said it many times when we first presented the UI to her!
19:29:54 [timbl_]Is it and easy fix?
19:30:02 [kennyluck]Yes, it is.
19:30:19 [timbl_]good
19:30:33 [timbl_]hmmmm ... I get now Error: YAHOO.namespace is not a function
19:30:34 [timbl_]Source File: chrome://tabulator/content/js/outline.js
19:30:34 [timbl_]Line: 2121
19:30:44 [timbl_]in the extension
19:30:47 [kennyluck]Q2: there are two kinds of input, as mentioned in http://dig.csail.mit.edu/issues/tabulator/msg576
19:31:13 [timbl_](in the extension when opening up Lydia's details)
19:31:25 [kennyluck]I know that issue.
19:31:41 [kennyluck]Conflict between my drag and drop code and birthday code.
19:32:37 [kennyluck]Q2 continued, the two kinds are a. autocomple on all predicates b. predicates matching rdfs:domain.
19:32:51 [jambo]timbl: quick fix: remove the display of a little calendar for birthdays in the outline
19:33:06 [kennyluck]Q2: should I merge these two into a menu?
19:33:23 [kennyluck]or enter->a click on selected TBD ->b?
19:34:53 [lkagal]lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
19:35:34 [timbl_]If the user clicks on something and anything happens apart from it getting selected or opened in some way, then the user is surprised.
19:36:06 [lkagal]lkagal has quit (Client Quit)
19:36:17 [timbl_]The normal way is to do a list wit the small number of predicates and have "more..." as one of the entries
19:36:31 [timbl_]or "more suggestions..."
19:36:57 [timbl_]keep that there on the bottom of the selection popup box
19:37:15 [kennyluck]Hold on, are you talking about (b)?
19:37:17 [jambo]kenny, let me make sure im getting this right
19:37:19 [timbl_]You need to make it obvious from the UI what will happen.
19:37:31 [jambo]you're wondering if hitting enter and click wait click should do the same thing?
19:37:44 [kennyluck]No.
19:37:58 [kennyluck]Just there are two kinds of predicate input as for now.
19:38:02 [timbl_]This is a new question
19:38:24 [kennyluck]I wonder if we should merge this two in a menu-submenu.
19:38:26 [timbl_]if i understand correctly, you have two possible auto-complete lists
19:38:31 [timbl_]One longer tham the other
19:38:43 [kennyluck]Yes.
19:39:11 [jambo]oh, i think i get it now
19:39:16 [timbl_]You are asking ho wto get from the small list to the bigger list
19:39:17 [jambo]you want to know if the user should get to pick?
19:39:28 [jambo]between small and big?
19:39:46 [jambo]or like if there should be a way to switch between them?
19:39:46 [kennyluck]sort of like that.
19:39:49 [timbl_]I suggest that you start withe the small list but give a button to switch it to the long list
19:39:58 [kennyluck]You should experience this a little bit.
19:40:10 [kennyluck]click the +
19:40:21 [timbl_]I have played with this.
19:40:26 [jambo]right
19:40:38 [kennyluck]try (a) type into the predicate (b) click the predicate when the TBD is selected.
19:40:42 [timbl_]I have often found that the autocomplete box is a few pixels wide and so it doesn't work properly.
19:40:58 [timbl_]Also I have found it is not in alphabetical order whcih makes it impossible to use
19:40:59 [jambo]kenny: that current display is really odd in windows: when i stop typing, the first one doesnt go away.
19:41:21 [jambo]kenny, have you looked at the way davidli's autocomplete works
19:41:41 [jambo]that is a good example of how it should probably behave, and then you can decide on how much you want to limit it
19:41:43 [timbl_]Also I have found that it doesn't any more remove the ones which don't match what I have types
19:41:46 [timbl_]typed.
19:42:12 [kennyluck]Are you referring to (a) or (b),timbl?
19:42:47 [jambo]kenny, i really recommend looking at davidli's autocomplete box. Short of limiting what symbols the user picks, it behaves in the proper way.
19:43:07 [timbl_]Well, normally with autocomplete, I type a few letters then from the now small list presented of things starting with that three, I click on one.
19:43:09 [kennyluck]Could you tell me what to do?
19:44:01 [kennyluck]timbl, that should remove the ones that don't match what you typed.
19:44:21 [timbl_]It does?
19:44:54 [timbl_]Right now I can't test it because I have a Error: tabStatusBar has no properties
19:44:55 [timbl_]Source File: chrome://tabulator/content/js/outline.js
19:44:55 [timbl_]Line: 1579
19:45:04 [timbl_]when I click onthe left-hand TBD.
19:45:37 [jambo]timbl, you updated 2005/ajar/ajaw but not 2007/tab
19:45:38 [timbl_]Tell you what to do> I think I have. Let me try again
19:45:40 [jambo]most likely.
19:46:07 [timbl_]1. When yo click on the plus, you shou dgo immediately to the selection of the predicate. Do not support the addition of first an object then a predictae.
19:46:30 [kennyluck]Ok.
19:46:44 [kennyluck]in this example, http://dig.csail.mit.edu/2007/wiki/test.rdf#test
19:47:13 [timbl_]That will mean you can go straight into the autocomplete mode for teh predicate, and the user will see what is happening.
19:47:40 [timbl_]2. There should be on the autocomplete box a "more suggestions" button which swicthes to the larger list.
19:47:47 [kennyluck]type into (TBD), or click that twice(non-consecutive) will show you two different menus.
19:48:03 [jambo]that should not happen kenny.
19:48:09 [jambo]it should always be the same menu
19:48:14 [timbl_]3. Theer should be a cancel x on autocomplete box which allows the user to back out of the whole add triple process, so it looks jsut like it did before they clicked "+"
19:48:23 [jambo]it's really really important that it's always the same menu
19:48:24 [timbl_](no residual TC=BD: TBD)
19:49:02 [kennyluck]hold on, I think there are two cases...
19:49:24 [kennyluck](a) the user knows what to input so he/she types.
19:49:41 [kennyluck](b)the user just got a new person and wonders what he/she can input.
19:49:59 [kennyluck]These are the two menus.
19:49:59 [jambo]kenny: it is always possible for you to take the user's input into the autocomplete box and not necessarily match it
19:50:11 [jambo]and reject it if you want to
19:50:19 [jambo]but you really should not have to separate types of menus
19:50:27 [jambo]that come from either "typing" or "clicking"
19:50:32 [jambo]that both provide the same functionality
19:50:49 [timbl_]jambo, you mean the two menus should not both exist, in different palces on the screen at the same time?
19:51:12 [jambo]well, thats really REALLY bad
19:51:25 [jambo]but no, i mean that those menus are two separate entities, and that is bad.
19:51:30 [timbl_]That is what I have seen
19:51:34 [jambo]if you want no autocomplete, you ought to be able to flag that on the same menu
19:51:53 [jambo]if you want limited autocomplete or every symbol ever in the store autocomplete, that should get flagged on the same menu
19:52:20 [jambo]right now there are two different menu elements getting defined for the same node's input. that is bad.
19:52:51 [timbl_]Jambo, I get that error only in the extension: var tabStatusBar = gBrowser.ownerDocument.getElementById("tabulator-display"); tabStatusBar has no properties
19:53:05 [jambo]timbl, are you sure you updated 2007/tab
19:53:17 [kennyluck]jambo, are you saying I should use menu-submenu?
19:53:18 [jambo]it sounds like you only updated 2005/ajar/ajaw
19:54:18 [timbl_]yes
19:54:45 [kennyluck]jambo?
19:54:49 [jambo]i'm not sure what menu-submenu means, kennyluck
19:55:05 [timbl_]So kenny, trying the "+" as it is, I click on the predicate TBD and I get a list with a scrollbar.
19:55:18 [timbl_]a) the list is not in alphabetical order
19:55:28 [jambo]in the interest of not overworking trying to put in new features, kennyluck, I would just go with an alphabetical list of all avail. preds
19:55:29 [timbl_]b) there is no text field for me to type in
19:55:57 [kennyluck]Just as I saide, there are two cases..
19:56:00 [jambo]and save making the list more specific for later
19:56:04 [jambo]no, no two cases.
19:56:12 [kennyluck]this is not a new feature and already exists there for a month.
19:56:14 [jambo]it really needs to be very straightforward
19:56:36 [jambo]there can't be two different menus.
19:56:49 [kennyluck]timbl, what you got is based on rdfs:domain.
19:57:10 [jambo]there can't be a "select one of these specific predicates" menu, and then some other "generic autocomplete" menu
19:57:31 [jambo]in terms of "implementing more features", i meant going to the trouble of making the single menu, and it is very important that this all happens on only _one_ menu
19:57:34 [timbl_]Jambo, how do you mean?
19:57:46 [timbl_]Where do you fidn these two menus
19:57:52 [timbl_]I only saw one
19:57:56 [jambo]they are the two that you were saying popped up for you !!
19:57:57 [kennyluck]...
19:58:42 [timbl_]Theer were when I tried before
19:58:45 [timbl_]Now I see one
19:58:50 [kennyluck]timbl, you don't seem to remember the first menu I showed you...
19:58:54 [jambo][15:47] <kennyluck> type into (TBD), or click that twice(non-consecutive) will show you two different menus.
19:58:55 [timbl_]Do you see the two problems above:
19:59:00 [jambo]that should not happen.
19:59:21 [timbl_]Agreed.
19:59:30 [timbl_]Clicking a second time should not do that
19:59:44 [timbl_]Kenny, don't try to be too smart.
19:59:53 [timbl_]Smarts have to be put in very carefully.
20:00:04 [jambo]and then, i said, that in the interest of simplicity, I thought that the first box (for the predicate) should just have all available predicates for this release
20:00:07 [jambo]with autocomplete
20:00:20 [jambo]and the second box should have all available symbols
20:00:30 [jambo]for the object
20:00:44 [kennyluck]kennyluck: (a) the user knows what to input so he/she types.
20:00:44 [kennyluck][3:49pm] kennyluck: (b)the user just got a new person and wonders what he/she can input.
20:00:53 [jambo]i'm suggesting this just for the sake of getting it out the door
20:01:07 [kennyluck]for new user who doesn't know ontology at all
20:01:16 [kennyluck]I think (b) is important
20:01:22 [jambo]having blank text on an autcomplete usually gives you the entire list
20:01:35 [jambo]once again, I suggest looking at davidli's autocomplete box
20:01:49 [kennyluck]I know, he didn't do (b).
20:02:05 [jambo]if the user is typing
20:02:07 [jambo]and has a blank string
20:02:12 [jambo]it shows all of the available symbols
20:02:17 [jambo]that is a ton of things that can be input
20:02:26 [timbl_]Kenny, when hte autocomplete box appears, the text field it applies to is not shown. I have to type a charecter and delete it to get the right effect. I have mentioned this bug lots of times I htink.
20:02:58 [kennyluck]the user would not want to input owl:disjointWith for a Person, jambo
20:03:08 [timbl_]I think the question of how to make a new node, like a new person, is inet=resteding.
20:03:27 [timbl_]However, I think we should ship this without it, and if we can make a nice solution, ship again
20:03:30 [jambo]kennyluck, but AIM_ChatID will still be in the list.
20:03:40 [timbl_]**** Release eary, release often ****
20:03:41 [jambo]so if the user starts typing AIM_ChatID, it will appear.
20:03:52 [timbl_]early
20:04:01 [jambo]if they start typing "Name", that will appear too
20:04:01 [kennyluck]look, a new users don't know there's a AIM_ChatID predicate there.
20:04:14 [jambo]kenny, this is for the sake of simplicity.
20:04:21 [kennyluck]You don't expect a user to find that predicate in the long list.
20:04:30 [jambo]no matter what you do, there will be a case where the user wants something else.
20:04:38 [timbl_]Kenny, why can't new users just scroll down the list and find AIM-chat ID?
20:04:42 [jambo]I'm telling you that for this release, i think it is more important that the predicates are there
20:04:45 [jambo]for the user to enter
20:05:04 [jambo]because the majority of the people that will be using it on this release will in fact know about a lot of predicates
20:05:19 [jambo]and if they dont, they will almost definitely be the type that can figure it out
20:05:24 [jambo]from a big long list
20:05:43 [jambo]and if they know something exists, but cant get to it, or its weird to get to it, they will be very annoyed that they cant enter the data that they want to
20:06:13 [jambo]it's not an easy problem to solve.
20:06:35 [kennyluck]Conclusion, eliminate the short list? I think this is sad.
20:07:14 [jambo]yes, i want you to eliminate the short list for this release just because it will make eliminating bugs for this release which is supposed to happen very very soon much easier.
20:07:45 [timbl_]The short list of of things which have a domain which explicitly matches a class of the object? That will exclude a lot of things
20:08:09 [jambo]especially considering right now we've had the bug pending "i get two menus when trying to edit" for well over a week
20:08:32 [timbl_]Kenny, for this release, please make just one menu.
20:09:03 [kennyluck]I guess I will do menu-submenu.
20:09:28 [jambo]kenny - what does that mean.
20:09:30 [kennyluck]It will do the user a flavor if the browser is doing a little bit of inference.
20:09:40 [timbl_]I know it is frustrating to suppress a creative bit of design you have done. But the art of getting this out is NOT the art of getting every nice feature in. So far, no one out there can use the tabulator as a browesr art all as an extension.
20:09:42 [kennyluck]favor..sorry.
20:09:43 [jambo]what does menu submenu mean
20:10:00 [timbl_]\I would be prepared to put out a release with NO editing capability if necessary, to ge the extension out.
20:10:10 [kennyluck]Suggested Prdicates -> another menu.
20:10:23 [jambo]kenny, only one menu
20:10:32 [jambo]please
20:10:35 [kennyluck]yeah, but submenu.
20:10:41 [jambo]that doesnt make any sense
20:10:49 [jambo]that is not how autocomplete menus behave
20:10:56 [timbl_]If it is easy to have a button to ask for more, like easier than going to one menu, then that would work too.
20:11:35 [kennyluck]I think we three have different opinions...
20:11:52 [jambo]what timbl is asking for is this:
20:12:32 [jambo] ______________________________
20:12:32 [jambo]| UserText_____ (more options) |
20:12:46 [jambo]or something to that effect
20:12:53 [jambo]with the autocomplete options below it
20:13:03 [jambo]that is something that i have seen before
20:13:05 [timbl_]kenny, note though that this inference the system is doing for the short list of not very solid. You can only really eliminate predicates whose domain is a class which is disjoint with a class you know the subject to be in. Later, we can work on getting it right, but really having just one menu -- with autoaomcplete -- will work find *for thi sreales*.
20:13:17 [jambo]right,
20:13:25 [timbl_]I think "more options' wouldbe best at the bottom of the list
20:13:32 [timbl_]as that is wher epeople expect more
20:13:59 [lkagal]lkagal (n=lkagal1@30-5-196.wireless.csail.mit.edu) has joined #dig
20:14:17 [jambo]the argument against that would be that its an expert user option and should therefore be quickly accessible (ie, what if there is a long list of options already available)
20:14:22 [timbl_]kenny, could you check in a version with just one menu, and work on thoer bugs?
20:14:23 [jambo]but that is not really important
20:14:28 [jambo]yes
20:14:42 [jambo]getting just one menu that works _very_ well is
20:14:49 [timbl_]After the release goes out, we can comment back in the fancier features and try them oiut more carefully.
20:15:38 [timbl_]=jambo, not at the bottom of the scolled area, below the scroll bar. But at the top is good too. ceratinly always visible
20:16:03 [timbl_]But it does have to work.
20:16:17 [jambo]oh, i see what you're saying now
20:16:26 [timbl_]Ahving the text field immediately selectde is important.
20:19:01 [kennyluck]By the way, when is "this" realease?
20:20:22 [timbl_]It is Jim's call, when we have enoug bugs out that it is basically working
20:20:48 [RalphS]RalphS has quit ("bye for today")
20:20:55 [kennyluck]Are you having release after that?
20:21:04 [kennyluck]we...sorry
20:21:18 [timbl_]:)
20:22:02 [jambo]kennyluck - "when its ready" or "at the end of summer"
20:22:34 [jambo]so "when its ready" or "sep 3"
20:23:06 [timbl_]It could be a very fast follow-up release in response to bugs found or user feedback, depending on how much energy and time we all put into it! (Term is coming!)
20:23:32 [jambo]with what will be an extremely unforgiving cut of all additional changes the week before sep 3 with the intention of making the release have only functional features, if it gets to be that late
20:23:38 [jambo];)
20:24:33 [timbl_]I am going to check in my changes, as the extension seems to work, both outline and table
20:25:18 [jambo]okay, i'll grind through everything for a half hour
20:25:57 [timbl_]darn, conflict in util.js
20:26:58 [presbrey]sorry, probably mine
20:27:19 [timbl_]no, there was no apparent edit in the svn change
20:27:26 [timbl_]maybe whitespace
20:27:27 [timbl_]ok
20:29:16 [timbl_]Committed revision 3718.
20:31:40 [timbl_]ie
20:32:10 [timbl_]Ok, example of a autocomplete bug.
20:32:20 [timbl_]Go to http://dig.csail.mit.edu/2007/wiki/Projects.rdf#test
20:32:21 [presbrey]darn, conflict in userinput.js
20:32:24 [jambo]timbl, really quick, did updating fix your problems with tabstatusbar ?
20:32:32 [timbl_]Press + under test.. name .. type. comment..
20:32:49 [timbl_]click on the predicate greed TBD
20:32:54 [timbl_]type d to get a text field
20:32:57 [timbl_]delete the d
20:33:27 [timbl_]click on "comment" in themenu,
20:33:42 [kennyluck]then?
20:33:52 [timbl_]... and the text field gets "_:n12" or some number in it.
20:34:05 [timbl_]Maybe I clciekd on a label for a predicate which is a bnode.
20:34:51 [timbl_]Now i can't re-create it
20:34:59 [kennyluck]I can't either...
20:35:20 [timbl_]Never mind then
20:35:55 [jambo]timbl, i've made a few calendars and timelines and such now with no errors, this seems like it works just fine
20:36:07 [timbl_]Good.
20:36:08 [kennyluck]Joe, what is the conflict?
20:36:14 [jambo]i'll be gone when you're back -- goodnight
20:36:34 [timbl_]night, you seem to be making progress.
20:36:36 [presbrey]Tim's changes; too near the sparql update changes I'm making
20:37:00 [timbl_]Thanks for helping fro taiwan, Kenny ... if you could fix the menu to be single that would be good, and fix the bugs yo know about.
20:37:25 [kennyluck]Uh-huh
20:37:27 [timbl_]My changes were pretty global, but should be obvious
20:37:31 [jambo]kenny, will you be available tomorrow
20:37:40 [kennyluck]should be.
20:38:03 [jambo]okay, don't worry about http://dig.csail.mit.edu/issues/tabulator/issue226 for now, we'll talk about that tomorrow
20:38:14 [timbl_]Presbrey, just using tabulator.ns.rdfs('label') instead of kb.sym('rdfs', 'label')
20:38:17 [jambo]and decide if it really matters
20:38:31 [jambo]byebye
20:38:42 [presbrey]yes its fine; just have to manually compute the diff for subversion
20:51:43 [jambo]jambo has quit ()
20:55:17 [presbrey]Kenny, I put placeholders for SPARQL error callbacks in userinput.js
20:55:46 [presbrey]you should update those with failure reactions
20:56:09 [presbrey]and disable editing until the callback returns
20:57:37 [kennyluck]Oh, Ok.
20:57:51 [presbrey](removed try/catch)
21:13:35 [davidli]davidli has quit ()
21:30:27 [apoorva]apoorva has quit (Read error: 110 (Connection timed out))
21:31:59 [presbrey]Kenny, adding blank node... dragging... can I drag between 2 windows?
21:34:04 [kennyluck]It used to work, I am not sure it's working now.
21:34:24 [kennyluck]you mean blank node?
21:34:32 [kennyluck]no, you can't.
21:34:53 [presbrey]you can't add a blank node?
21:34:58 [presbrey]how do you add based_near with lat/long?
21:35:23 [kennyluck]I mean, you can't drag blank node across windows.
21:37:09 [presbrey]how can I add a based_near lat and long to wiki/test#test
21:39:33 [kennyluck]No simple method for now, sorry.
21:40:10 [kennyluck]What you can do is to cheat.
21:40:49 [kennyluck]By the way, we haven't had a agreement on how to edit bnode like that.
21:41:27 [kennyluck]Is based near a FP?
21:42:35 [kennyluck]hold on,
21:42:46 [kennyluck]based-near is a ObjectProperty.
21:43:07 [kennyluck]huh...I'm sorry it does not help for now.
21:55:43 [lkagal]lkagal has quit ()
21:59:06 [presbrey]your labeler method was broken
21:59:16 [presbrey]never produced a where clause for updating
21:59:32 [presbrey]so I replaced it back with the old one
21:59:35 [presbrey]in prepareUpdate
22:02:07 [kennyluck]How could that be? Did you try to update a bnode?
22:02:39 [presbrey]yes I edited jim's based_near
22:03:26 [kennyluck]Oh, in that case, maybe your algorithm is better.
22:03:42 [kennyluck]but for example, this is your code s = this.store.statementsMatching(undefined,undefined,s.subject,statement.why);
22:04:01 [kennyluck]a knows bnode1. b knows bnode2.
22:04:42 [kennyluck]even though s.length for statementsMatching(undefined,undefined,bnode1,why) is 1
22:05:10 [kennyluck]but WHERE {a knows ?bnode} does not identify bnode1
22:05:29 [kennyluck]sorry, my example was. a knows bondes1. a knows bnode2.
22:07:07 [kennyluck]I think my code may have this problem as well, I'm not sure.
22:09:15 [presbrey]it depends what you're editing
22:10:08 [presbrey]if you know two people named Joe and you try to change one of their names to Joseph instead, it currently fails because there it doesn't do anything with IFPs
22:10:51 [presbrey]if you have two blank nodes that are distinct based_near's with lats and you change 1 lat, it works
22:11:09 [presbrey]dig.csail appears down
22:11:57 [kennyluck]But in that Joe case, with my code, s.length=2 and will not give a WHERE (should throw an error)
22:12:53 [presbrey]something else is wrong with your code
22:13:03 [presbrey]it never produces a where clause for *any* example
22:13:34 [kennyluck]It did. Lots of wiki-logs are produced by my code.
22:15:02 [presbrey]I looked at the logs; there hasn't been a where since 8/2
22:15:25 [kennyluck]look at those Dennis cases...
22:17:57 [kennyluck]say, 8/2
22:18:13 [kennyluck]WHERE { ?n73 <http://xmlns.com/foaf/0.1/name> "name: foo" . }
22:18:13 [kennyluck]DELETE { ?n73 <http://xmlns.com/foaf/0.1/name> "name: foo" . }
22:18:41 [kennyluck]?n73 foaf:name "name:foo" is an entry in the labeler
22:19:54 [kennyluck]it's sad that test.rdf is screwed up while I was testing DELETEs.
22:20:11 [kennyluck]we don't have many examples to play for now.
22:22:55 [timbl_]Anyone at MIT?
22:23:21 [timbl_]Ah, and you have been introduced to Mark Pearrow and garret.
22:23:43 [timbl_]You could zoom down and see if they are there messing with groups.csail.mit.edu
22:24:05 [presbrey]yes I will go complain on my way out
22:24:35 [timbl_]It would be good to know what has been going on, and find out whether they have been awre of it.
22:24:39 [timbl_]Thanks in advance!
22:25:03 [presbrey]Kenny, it looks like there's already a User:Kennyluck
22:25:16 [presbrey]was it not working?
22:25:26 [kennyluck]I meant to user User:Kennyluck but not User:Kennyluck.rdf
22:26:15 [presbrey]did you try without the ".rdf"?
22:26:36 [presbrey]I'll go now... I can't use tabulator without dig =\
22:27:01 [kennyluck]I got Unable to connect.
22:28:08 [kennyluck]By the way, I did an a-line hack to your _get.php so 404 is served with content-type: application/rdf+xml
22:28:30 [presbrey]yeah I saw that
22:29:04 [presbrey]hmm I might get locked out going down there... its late
22:29:56 [presbrey]presbrey has quit ()
22:43:03 [kennyluck]what's wrong with the current revision. I got lots of Errors?
22:49:25 [presbrey]presbrey (n=presbrey@c-24-61-47-67.hsd1.ma.comcast.net) has joined #dig
22:50:38 [kennyluck]good job, Joe.
22:53:32 [timbl_]Errors?
22:54:14 [kennyluck]So Joe, I type in http://dig.csail.mit.edu/2007/wiki/User:Kennyluck# and I get 404 in HTML
22:54:40 [kennyluck]david's incomplete sparqlUpdate code, it seems.
23:09:23 [timbl_]The server seems to be broken now. Not completely down but in a mess.
23:10:03 [kennyluck]sometimes it directs to localhost//afs ?? what is this? webDav?
23:22:22 [presbrey]User:Kennyluck should work now
23:23:31 [kennyluck]It works. Nice!
23:24:42 [timbl_]Joe, what was up?
23:25:41 [presbrey]I had limited the wiki to use filenames with extensions ".rdf" or ".n3"
23:25:47 [presbrey]just took it out is all
23:26:19 [timbl_]Are they in fact stored as rdf/xml always?
23:27:01 [timbl_]It is wise to add a .rdf in the filesystem but suppress it from the URI users see
23:27:09 [presbrey]yes, limited by what algae will work with
23:27:39 [timbl_]Is the webdav working too nowadays?
23:27:45 [presbrey]you should put something in the wiki in tabulator's default root
23:28:24 [presbrey]your smushing/serialization problem
23:28:29 [presbrey]prevents use of webdav
23:28:44 [kennyluck]Joe, User:Kennyluck still gives me 404 even there's data in it.
23:29:13 [presbrey]no, I get 200
23:29:35 [kennyluck]Ah, sorry. I get 200 now.
23:29:44 [presbrey]always try curl first
23:29:45 [timbl_]smushing/serialization ... you mean that the exact statements aren't available when smushing happens ... yes .. that I am not supposed to touch that isseu till after the release.\
23:30:11 [presbrey]so webdav will have to wait then, Tim
23:31:37 [timbl_]I find it difficult to estimate when this release will happen!
23:32:15 [presbrey]I rewrote much of sparql update today; it had been rather botched. it works much better.
23:32:34 [presbrey]I don't see many other issues on the list
23:33:14 [presbrey]that aren't already taken
23:33:25 [presbrey]and blocking a release
23:33:26 [presbrey]whats up?
23:50:06 [presbrey]issue tracker is down
23:50:24 [timbl_]Dod you see the folks on the 2nd floor?
23:51:05 [timbl_]Yes, 404
23:51:06 [presbrey]no
23:51:26 [timbl_]They say they haven't switched the server yet
23:51:33 [timbl_]but anyone would think they had
23:52:27 [timbl_]They have problems finding out how to port OpenID
23:53:19 [presbrey]how is that related to running the dig site?
23:53:51 [timbl_]OpenID is used for the dig blog
23:53:58 [timbl_]Ryan installed it
23:54:15 [timbl_]They triedto clone the site and couldn't apt-get install openid
23:57:46 [presbrey]well let me know if you remember any of the things that were on the tracker that I can do
23:57:51 [presbrey]presbrey has quit ()