IRC log of sparql-func on 2009-09-15

Timestamps are in UTC.

13:05:22 [RRSAgent]
RRSAgent has joined #sparql-func
13:05:22 [RRSAgent]
logging to http://www.w3.org/2009/09/15-sparql-func-irc
13:07:00 [AxelPolleres]
paul, do you plan to dial in still?
13:07:42 [pgearon]
yes, just dialling now
13:08:01 [AndyS]
-> http://www.w3.org/2009/sparql/wiki/Feature:FunctionLibrary
13:08:03 [Zakim]
+pgearon
13:08:44 [AxelPolleres]
http://www.w3.org/2009/sparql/wiki/Feature:FunctionLibrary#Starting_Points
13:09:08 [kasei]
kasei has joined #sparql-func
13:10:22 [AxelPolleres]
lee, you are very hard to understand
13:10:32 [LeeF]
I'll type
13:10:48 [LeeF]
low hanging fruit for me is things that already exist, are implemented, and use data types within sparql
13:10:54 [LeeF]
full-text search is def. not low hanging fruit to me
13:11:02 [LeeF]
coalesce/if is on the boundary of what i'd want to consider, it scares me a bit
13:11:06 [LeeF]
(but is very useful)
13:13:07 [AxelPolleres]
andy: getNamespace, getlocalname are pragmatic but have proven to be confusing.
13:13:07 [LeeF]
I wonder if we should discuss "classes" of functions? String functions, Date/datetime functions, Math functions, URI functions, logic functions, SPARQL-specific functions ?
13:14:29 [AndyS]
Good idea.
13:14:38 [AxelPolleres]
Andy: we should define agreed core with a chance to be implemented widely.
13:15:08 [LeeF]
I don't necessarily agree to that. I'd prefer not to add a lot of keywords
13:15:15 [AxelPolleres]
keywords vs. iris
13:16:14 [LeeF]
I don't think we'll be able to pull whole function libraries in, I think we'll have to be selective...
13:16:28 [AndyS]
I had: Specials (IF) // Constructed e.g. String to IRI // Library choices and library can be string, math, date/time etc
13:17:19 [AxelPolleres]
AndyS: keywords only where we need to.
13:17:28 [LeeF]
constructors
13:17:36 [AxelPolleres]
turning strings into IRIs would be useful.
13:18:08 [LeeF]
I think we should try to pick functions from just one library if possible
13:18:24 [AndyS]
SQL has different evaluation
13:18:27 [pgearon]
I don't see overlap as a problem, so long as namespaces are used
13:18:42 [LeeF]
I think it's a little silly to standardize two substring (or whatever) functions
13:19:41 [AxelPolleres]
Our take on overlaps would be to pick IRIs, not overlapping.
13:20:14 [AxelPolleres]
shall we also pick a namespace for the existing ones?
13:21:08 [AxelPolleres]
PROPOSED: pick one namespace for the core functions.
13:21:45 [LeeF]
in Glitter I use "http://www.w3.org/2006/sparql-functions#" -- did I invent that or is that codified somewhere?
13:22:50 [AndyS]
Invent I think but we coudl use that (rewrite history!)
13:23:13 [pgearon]
LeeF, it gives a 404 on the W3C, so I suspect that it's not official
13:23:41 [LeeF]
I must have seen it somewhere, since why would I have invented the /2006/ part of it?
13:23:41 [LeeF]
hmm
13:23:48 [AxelPolleres]
sqrt?
13:23:56 [pgearon]
My main problem with XPath/XQuery is finding a library that implements it! :-)
13:24:08 [LeeF]
http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/att-0138/sparql-function-uri.html
13:24:28 [AxelPolleres]
Agreement is to start with a subset of xpath/xquery.
13:24:30 [AxelPolleres]
?
13:24:34 [LeeF]
(and http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0168.html )
13:24:40 [LeeF]
I'm happy with that.
13:25:02 [AndyS]
+1
13:26:09 [AxelPolleres]
http://www.w3.org/TR/rif-dtb/
13:27:41 [AxelPolleres]
no exact different error values in SPARQL
13:28:56 [AndyS]
&& and ||
13:30:14 [AxelPolleres]
Andy: our error is like a "superclass" of all the xpath/xquery errors
13:31:37 [AndyS]
We can ask for me from XQ
13:31:43 [AndyS]
We can ask for one from XQ
13:32:17 [pgearon]
heh
13:32:35 [pgearon]
should we create a namespace for "op:"
13:33:12 [LeeF]
i play a bit dirty in Glitter:
13:33:13 [LeeF]
static public final String OP_NAMESPACE = "http://www.w3.org/2005/xpath-functions#";
13:33:19 [AxelPolleres]
Option 1: base on XPath/XQuery and invent a namespace only for those which are separate.
13:33:34 [AxelPolleres]
Option 2: take our own namespace for all
13:33:54 [AxelPolleres]
Option 1 caveat: no namespace for op:
13:35:45 [AxelPolleres]
?X > ?Y
13:36:58 [AxelPolleres]
is problematic... since it can boil down to different op: functions depending on binding of ?X ?Y
13:36:58 [AndyS]
LeeF - I think I do same but I'm on the wrong machine to check.
13:37:29 [AxelPolleres]
sparql:less-than
13:37:35 [AxelPolleres]
>
13:38:14 [pgearon]
While I'm quite happy to duplicate existing functionality in various namespaces, I don't see it as mandatory
13:38:18 [LeeF]
Andy - maybe I copied you :)
13:38:38 [LeeF]
Why do we define URIs for the existing operators?
13:38:44 [LeeF]
s/do we/do we need to
13:39:00 [pgearon]
LeeF, for completeness.... but I don't see it as essential
13:39:11 [AndyS]
-> http://www.w3.org/TR/rdf-sparql-query/#OperatorMapping
13:39:18 [pgearon]
it's basically overengineering :-) (But the W3C is good for that)
13:39:18 [LeeF]
completeness is not a big (personal) motivator to me :)
13:39:40 [AndyS]
It would nice because SPARQL is dynamic to have a IRI for each symbol in that table.
13:39:44 [AndyS]
:-)
13:39:57 [pgearon]
but that's my point... it would be "nice"
13:40:01 [pgearon]
not essential
13:40:04 [LeeF]
right, but.... so? :-)
13:40:13 [LeeF]
right!
13:40:16 [LeeF]
not going to do that :)
13:40:23 [LeeF]
?foo(?bar)
13:40:25 [AndyS]
?var(?a, ?b, ?c)
13:40:28 [pgearon]
I'd like it, but I don't see it as essential
13:40:41 [LeeF]
no no, not interesting at all :)
13:40:47 [pgearon]
Andy: That'd be so cool
13:40:56 [pgearon]
I've already been talking with people about doing that in Mulgara :-)
13:40:56 [AxelPolleres]
-1 to that... looks suspicious.
13:41:07 [LeeF]
pgearon, stop that
13:41:08 [LeeF]
:)
13:41:16 [AndyS]
eval(?func, ?a, ?,?c)
13:41:28 [pgearon]
LeeF, I'm not kidding. Have been seriously considering
13:41:53 [LeeF]
yes, that's the part that worries me
13:42:02 [LeeF]
;)
13:42:42 [AxelPolleres]
In RIF we have that... special keyword External( ... ) to delimit externally evaluated functions/predicates
13:42:56 [pgearon]
I don't know that eval is needed. So long at the namespace for an "unknown" function leads me to something
13:44:48 [AxelPolleres]
ISSUE: shall we allow variables in function position?
13:45:19 [LeeF]
I'm very much against that, seems way beyond the scope of what we set out to do
13:45:22 [pgearon]
just don't forbid it, and I can work with it as a valid extension :-)
13:45:35 [LeeF]
No, no agreement there
13:45:38 [LeeF]
IF might need one
13:45:42 [AndyS]
-1
13:45:46 [LeeF]
IF might need one
13:45:51 [AndyS]
-1 to ?var()
13:46:09 [pgearon]
I don't want to see a new keyword. Just a namespace that must (?) be implemented
13:46:38 [pgearon]
something like sparql:
13:46:43 [LeeF]
pgearon +1
13:46:54 [pgearon]
duplicate functionality into that namespace
13:47:32 [AxelPolleres]
so, we put ALL core functions into the core namespace?
13:48:31 [AxelPolleres]
+1 to keep all core in our namespace.
13:49:05 [AxelPolleres]
paul: sparql functionality in terms of 3 different namespaces sounds messy.
13:49:47 [LeeF]
I guess I don't really care about the URIs. I'm much more interested in which functions we adopt
13:49:52 [AxelPolleres]
... the one namespace could be the standard namespace.
13:49:57 [pgearon]
"defined as part of the core library". I like that way of putting it
13:50:08 [AxelPolleres]
AndyS: we already allow other namespaces.
13:50:13 [AndyS]
-1 to namespace : +1 to LeeF
13:50:18 [pgearon]
just a reminder: we're 10 minutes out from the SPARQL call
13:51:28 [AxelPolleres]
let's continue at this slot next or in two weeks?
13:51:28 [pgearon]
I won't be available next week, but fine for 2 weeks
13:51:36 [AxelPolleres]
I will try to summarize the status.
13:51:46 [AndyS]
What actions will eb competed by then?
13:52:43 [AxelPolleres]
ACTION: Axel to look over andy's core set.
13:53:07 [LeeF]
AndyS++++
13:53:20 [pgearon]
+1 for Andy's suggestion of defining a set of functions (eg. 5 string functions)
13:53:40 [pgearon]
This is orthogonal to the namespace issue. I'm happy with it
13:53:51 [AxelPolleres]
Andy: my interest in the libary is that it is available everywhere. small set that everyone has would be a big step forward.
13:54:02 [AxelPolleres]
+1
13:54:31 [pgearon]
Looks like we have consensus on defining a set of functions
13:55:29 [AxelPolleres]
ACTION: to mail result of previous acttion to list and ask who supports what
13:55:41 [Zakim]
-LeeF
13:55:43 [pgearon]
Thanks everyone
13:55:45 [Zakim]
-AxelPolleres
13:55:47 [kasei]
kasei has left #sparql-func
13:55:48 [Zakim]
-pgearon
13:55:58 [Zakim]
-AndyS
13:56:00 [Zakim]
Team_(sparql-func)13:02Z has ended
13:56:01 [Zakim]
Attendees were AndyS, +3539149aaaa, AxelPolleres, LeeF, pgearon
13:56:45 [AxelPolleres]
THanks!
13:56:45 [AxelPolleres]
BTW: anyone knows *where* these actions are tracked or whether they are at all?
13:57:01 [AxelPolleres]
I assume I can use the normal SPARQL action tracker?
13:58:00 [AxelPolleres]
rrsagent, make records public
13:58:12 [AndyS]
RRSagent puts them in the minutes
13:58:22 [AndyS]
trackbot is the wiki thing
13:58:26 [LeeF]
If you want to use the regular sparql tracker, you'll need to port them yourself
14:00:04 [AndyS]
AndyS has left #sparql-func
14:02:36 [AxelPolleres]
yup
14:05:06 [AxelPolleres]
:-) thanks all!
14:06:40 [AxelPolleres]
RIF: Axel still following up, LC comments closed, going to Cand. Rec ... ah, ivan reporting already ;-)
14:07:20 [AxelPolleres]
wrong window, oops. ignore.
14:14:37 [pgearon]
pgearon has left #sparql-func
14:19:19 [AxelPolleres]
member:rrsagent, make public minutes
14:19:29 [AxelPolleres]
rrsagent, make public minutes
14:19:29 [RRSAgent]
I'm logging. I don't understand 'make public minutes', AxelPolleres. Try /msg RRSAgent help
14:19:32 [AxelPolleres]
AxelPolleres has left #sparql-func
15:32:00 [Zakim]
Zakim has left #sparql-func