IRC log of dig on 2010-10-31

Timestamps are in UTC.

00:17:59 [presbrey]

good evening everyone

00:18:29 [presbrey]

timbl, in Tabulator, how do I create a new RDF document from empty?

00:18:36 [melvster]

melvster has quit (Ping timeout: 240 seconds)

00:23:21 [presbrey]

if I curl a few triples up, its easy to edit them

00:23:38 [presbrey]

but if I'm starting from nothing in Tabulator, I'm not sure how to insert a new statement

00:28:26 [presbrey]

ericP, I have everything working in PHP but swobjects

00:28:46 [presbrey]

I will hook up application/sparql-query POST to python for now

00:29:03 [presbrey]

but it would be nice to trim these dependencies a bit

01:21:28 [jnpato_]

jnpato_ has quit (Ping timeout: 252 seconds)

01:43:32 [presbrey]

SWObjects base URI support is not working right

01:43:36 [presbrey]

when you're around ericP

01:47:13 [presbrey]

looks like RdfDb.toString does not use base URI

01:55:20 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

01:57:11 [lkagal]

lkagal has quit (Client Quit)

02:13:41 [kennyluck]

kennyluck (~kennyluck@EM114-48-111-215.pool.e-mobile.ne.jp) has joined #dig

02:23:36 [kennyluck]

kennyluck has quit (Ping timeout: 240 seconds)

04:07:25 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

04:50:40 [ericP]

presbrey, i bounced the link prob off a friend and he said he doesn't think the E_NOTICE is disabling dynamically loaded extensions, but instead some configuration

04:51:34 [ericP]

[[

04:51:36 [ericP]

PHP Notice: Constant swig_runtime_data_type_pointer already defined in Unknown on line 0

04:51:57 [ericP]

PHP Warning: dl(): Dynamically loaded extensions aren't enabled in t_SWObjects.php on line 3

04:52:00 [ericP]

]]

05:25:48 [lkagal]

lkagal has quit (Quit: lkagal)

06:21:01 [nunnun]

nunnun is now known as nunnun_away

07:30:46 [nunnun_away]

nunnun_away is now known as nunnun

07:34:21 [nunnun]

nunnun is now known as nunnun_away

08:51:21 [nunnun_away]

nunnun_away is now known as nunnun

09:31:30 [nunnun]

nunnun is now known as nunnun_away

09:37:15 [nunnun_away]

nunnun_away is now known as nunnun

11:42:00 [timbl]

presbrey, Good question.

12:10:40 [melvster]

melvster (~melvster@p5797FD46.dip.t-dialin.net) has joined #dig

13:08:58 [melvster]

melvster has quit (Ping timeout: 265 seconds)

13:29:17 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

13:31:45 [ericP]

presbrey, could you give me a curl invocation that emulates the POST with application/sparql-query ?

13:37:50 [timbl]

See the DesignIssues

13:43:04 [ericP]

nice!

13:44:38 [lkagal]

lkagal has quit (Quit: lkagal)

14:11:39 [presbrey]

good morning

14:12:27 [presbrey]

curl --cert ./webid.pem -kv -H 'Content-type: application/sparql-query' --data 'INSERT DATA { <a> <b> <c> } WHERE {}' https://presbrey.rdf.me/test/insert

14:14:50 [presbrey]

ericP, you shouldn't use dl with t_SWObjects

14:15:23 [presbrey]

when you call php, use `php -d extension=swobjects.so t_SWObjects.php`

14:16:20 [presbrey]

dl is missing/disabled from many phps now

14:16:50 [presbrey]

if its loaded before-hand with extension=, the swig wrapper should skip the dl load code

14:18:44 [ericP]

does that fix the php on the dig box?

14:18:49 [ericP]

s/box/vm/

14:19:06 [presbrey]

maybe, I'll check

14:19:51 [presbrey]

I'm also having a problem getting toString to serialize with relative URIs

14:20:28 [presbrey]

or I forgot the magic switch

14:20:32 [ericP]

it works in python?

14:20:38 [ericP]

or is this a general bug...

14:20:41 [presbrey]

I'm not sure

14:21:10 [presbrey]

in python it doesn't work

14:21:22 [presbrey]

I'm using python on rdf.me now

14:21:54 [presbrey]

eg. SPARQL POST: INSERT DATA { <a> <b> <c> } WHERE {} to https://presbrey.rdf.me/test/insert

14:22:21 [presbrey]

results in seeing https://presbrey.rdf.me/test/insert when I request eg. http://presbrey.rdf.me/test/insert.json

14:23:25 [ericP]

can you add a test for it in t_SWObjects.py ?

14:23:31 [ericP]

(and commit)

14:23:47 [ericP]

(so's i can work an a machine with a litttle more space)

14:24:25 [presbrey]

sure

14:25:03 [ericP]

tx

14:25:28 [presbrey]

all negotiations have it, eg. http://presbrey.rdf.me/test/insert.nt

14:25:47 [presbrey]

vs. POST Turtle (append) works fine

14:25:52 [presbrey]

but that doesn't use SWObjects

14:26:22 [presbrey]

toString doesn't take a base URI, does it?

14:26:28 [presbrey]

where am I supposed to set that, RdfDB?

14:26:59 [ericP]

i think it does take a baseURI

14:27:31 [ericP]

hmm

14:27:38 [presbrey]

I looked in RdfDB.{c,h}pp

14:28:05 [ericP]

std::string BasicGraphPattern::toString (MediaType mediaType, NamespaceMap* namespaces) const

14:28:20 [ericP]

lib/SWObjects.cpp:1866

14:28:33 [ericP]

not much visibility of a baseURI there

14:28:43 [presbrey]

no

14:28:56 [presbrey]

so maybe only the RDF/SPARQL parsers are supposed to support it so far

14:28:57 [ericP]

what's the natural calling convention there?

14:29:23 [presbrey]

a serializer will take a base URI argument

14:29:24 [ericP]

yeah, it works on the input, just doesn't round trip

14:29:43 [presbrey]

use the base URI of its bound model

14:29:49 [presbrey]

or allow an override passed

14:30:19 [presbrey]

and it should be before namespaces

14:30:39 [ericP]

we have

14:30:42 [ericP]

bgp->toString(MediaType("text/turtle"), namespaceMap);

14:30:47 [ericP]

we'd like:

14:30:58 [ericP]

bgp->toString(MediaType("text/turtle"), baseURI, namespaceMap); ?

14:31:03 [ericP]

bgp->toString(MediaType("text/turtle"), namespaceMap, baseURI); ?

14:31:27 [presbrey]

I like the former

14:32:00 [presbrey]

baseURI and namespaceMap should be nullable

14:32:22 [ericP]

bgp doesn't store stuff like the current baseURI or namespaces (when constructed by the parser), but i guess it could

14:32:42 [ericP]

but for now, i'll just make the caller keep track of it: bgp->toString(MediaType("text/turtle"), baseURI, namespaceMap);

14:32:43 [presbrey]

doesn't need to now

14:34:06 [ericP]

c++ is fussy (probably unnecessarily so) about optional parms; once you NULL one, you can't supply an more

14:35:05 [ericP]

s/NULL/default/

14:35:33 [presbrey]

you can't typically skip optional arguments

14:35:54 [presbrey]

Python lets you if you specify the optional argument you are skipping to by name

14:36:01 [ericP]

gotcha, which means that we'd need both:

14:36:07 [ericP]

toString(MediaType type = MediaType("text/turtle"), const TTerm* baseURI = NULL, NamespaceMap map = NamespaceMap());

14:36:22 [ericP]

toString(MediaType type = MediaType("text/turtle"), NamespaceMap map);

14:36:48 [presbrey]

why?

14:37:11 [presbrey]

why not always explicitly send a NULL baseURI?

14:37:16 [ericP]

so you don't have to add a NULL to every toString which currently supplies a NamespaceNap

14:37:22 [ericP]

ahh

14:37:35 [ericP]

maybe that's the right level of in-your-face

14:37:37 [presbrey]

it seems bad not to send a baseURI, I would make people be conscious of it

14:38:17 [ericP]

as a computitarian, i feel like the typing SHOULD figure all that out for you, IF the types are unambiguous

14:38:24 [ericP]

but yeah, your API makes sense

14:39:08 [ericP]

hhm, can you think of any places where i do relative uri resolution?

14:39:44 [ericP]

i guess that's the goal of the HT_URI code i still from libwww (and still don't know how to use)

14:40:14 [ericP]

ht_parse, i guess

14:41:25 [ericP]

actually, can you write me a nice recipe (in whatever language, though C++ using std::string would make it paste-able) which takes a URI string and an optional base and "does the right thing"?

14:41:42 [presbrey]

I was going to spend the next few minutes doing the tabulator auto-update RDF manifest

14:42:01 [presbrey]

right thing in terms of outputting a relative uri?

14:42:26 [ericP]

yeah, including not absolutizing behind what the base uri suggests

14:43:01 [presbrey]

behind? eg. I don't think the current relativizer puts in ../../ in cases

14:43:38 [ericP]

yeah, let's see what we'd like:

14:43:40 [ericP]

base("foo/") + "bar" => "foo/bar"

14:43:50 [ericP]

base("foo/") + "../bar" => "bar"

14:44:15 [ericP]

base("../foo/") + "bar" => "../foo/bar"

14:44:39 [ericP]

base("../foo/") + "../bar" => "../bar" # i think

14:44:53 [ericP]

base("/foo/") + "bar" => "/foo/bar"

14:45:01 [presbrey]

base("/foo/bar/") + "/foo/" => "../"

14:45:16 [ericP]

ahhh, interesting

14:45:33 [presbrey]

base("/foo/bar/") + "/foo/bar/" => "."

14:46:09 [ericP]

rule (guideline) appears to be "always keep it as relative as possible"

14:47:42 [oshani]

oshani (~oshani@31-33-76.wireless.csail.mit.edu) has joined #dig

14:48:20 [ericP]

i don't think 2396 licenses you to include the fqdn but elide the protocol a la: base("example.com/foo/") + "bar" => "example.com/foo/bar"

14:48:24 [ericP]

but that would be cool too

14:48:40 [ericP]

(would work for e.g. ftp and http)

14:49:24 [presbrey]

sure

14:51:03 [presbrey]

here's a version in JS: http://github.com/mootools/mootools-more/blob/master/Source/Types/URI.Relative.js

14:51:09 [ericP]

hmm, need to extend to anchors as well

14:51:27 [presbrey]

look at combine: function(bits, base)

14:51:28 [timbl]

(presbrey, Ok, so where should there be a UI for adding completely new dat to an empty file?)dig-hg/tabulator/chrome/content/test/tc0000

14:51:30 [ericP]

interruping cow?

14:51:59 [ericP]

presbrey, does mootools do all the erring on the side of relative that you'd like?

14:52:12 [presbrey]

ericP, yes

14:52:20 [timbl]

tabulator/chrome/content/test/tc0000/test_uri.js

14:52:24 [presbrey]

timbl, might be simple as putting a (+) button somewhere

14:53:02 [timbl]

Well, then have to think of whether to make up a new random URI like <#a> or ask the user for one

14:53:27 [presbrey]

why not put in a fillable-blank

14:53:32 [timbl]

but the questionw as where.

14:53:56 [ericP]

crikey, that's short

14:53:57 [timbl]

Maybe a new little pane which comes up for any empty RDF file.

14:53:58 [presbrey]

can't blanks be disambiguated explicitly with an id?

14:54:05 [ericP]

(the .js)

14:54:23 [ericP]

indicates that found *the* *thruth*

14:54:29 [ericP]

s/that/they/

14:55:51 [timbl]

they?

14:55:54 [presbrey]

timbl, the pane should come up automatically for a 404

14:56:03 [presbrey]

when the content-type is RDF-like

14:57:46 [presbrey]

timbl, eg. I think tabulator should edit http://test.rdf.me/empty

14:58:18 [timbl]

file:///devel/dig-hg/tabulator/chrome/content/test/tc0000/index.html

14:58:39 [presbrey]

ahh, absolute URI!

14:58:56 [timbl]

That isb't an empty file. That is a 404 nonexistant file.

14:59:11 [presbrey]

did you curl it?

14:59:22 [presbrey]

it has an <rdf />

14:59:27 [tlr]

tlr (~tlr@89.204.137.254) has joined #dig

14:59:34 [presbrey]

it has an MS-Author-Via: SPARQL

14:59:38 [timbl]

No, I looked it up in Tabr, and looked in the Internetals pane , clocied on the URI and the request and the response

14:59:43 [presbrey]

it has a Content-Type: text/turtle

15:00:13 [timbl]

But at 14:58:05.215 the server said 404

15:01:14 [presbrey]

at 14:58:05, the server said "this graph URI isn't found but you can post SPARQL to create it"

15:01:41 [presbrey]

(you can also post turtle or rdf/xml to create it)

15:02:00 [timbl]

No, to do that it would have to send back 200

15:02:02 [timbl]

$ curl -I http://test.rdf.me/empty

15:02:02 [timbl]

HTTP/1.1 404 Not Found

15:02:02 [timbl]

Date: Sun, 31 Oct 2010 15:01:16 GMT

15:02:02 [timbl]

Server: Apache

15:02:03 [timbl]

Content-Type: text/html; charset=iso-8859-1

15:02:17 [presbrey]

don't use -I, that sends HEAD and HEAD is broken right now

15:02:24 [presbrey]

do curl -v http://test.rdf.me/empty

15:02:41 [timbl]

the datawikie send back content-type application/rdf+xml and a vaid file with no triples in it

15:02:48 [tlr]

tlr has quit (Client Quit)

15:04:06 [presbrey]

there are at least 3 kinds of graphs in terms of existence

15:04:34 [timbl]

Ah, Ok, so HEAD is just broken.

15:04:34 [presbrey]

a) existing graph containing >=1 triple: graph <test> { <a> <b> <c> }

15:04:45 [presbrey]

a) existing graph containing no triples: graph <test> {}

15:05:00 [presbrey]

c) non-existing graph: (null)

15:05:18 [presbrey]

I argue only (a) and (b) should return 200, (c) should return 404

15:05:48 [timbl]

Well, it depends on whether you are doing the eiki thing, which is to assume everuything exists.

15:06:23 [presbrey]

why can't we do the wiki thing another way?

15:06:35 [timbl]

So in a wiki everything is 200, like wikipedia

15:06:59 [presbrey]

=> if a server tells you it supports SPARQL during a 404, you can wiki/sparql/post-append create that document/graph URI

15:08:09 [presbrey]

what I suggest is the real way wikipedia works, for example.

15:08:13 [presbrey]

and it makes sense to me as well

15:08:17 [timbl]

Actually I lie. The wikipedia http://en.wikipedia.org/wiki/RDFalkjlsajfldjflgkjSKDLFALKJLKASDJFKLASD returns 404 but a nice 4040 page whcih looks like a uccess

15:08:24 [timbl]

success

15:08:46 [presbrey]

yes, the former way you suggest is more like comcast/roadrunner always replying DNS entries when they don't exist

15:08:57 [timbl]

yes.

15:09:17 [presbrey]

lets not do liken ourselves to that

15:09:28 [timbl]

Yes.

15:09:40 [presbrey]

~]$ curl -v http://en.wikipedia.org/wiki/Joe_Presbrey 2>&1 | grep 'HTTP/1'

15:09:40 [presbrey]

> GET /wiki/Joe_Presbrey HTTP/1.1

15:09:41 [presbrey]

< HTTP/1.0 404 Not Found

15:10:13 [timbl]

ok, so we then need to have a way to allow the system to bootstrap itself up to editing a new file.

15:10:37 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

15:10:50 [timbl]

What should the tabulator do first to use a new file? A PUT?

15:11:02 [timbl]

(of a zero-triple file)

15:11:30 [presbrey]

a POST of zero-triples

15:11:45 [presbrey]

or a SPARQL INSERT DATA {} WHERE {}

15:12:28 [presbrey]

PUT/DAV should be avoided whenever possible

15:12:40 [lkagal]

lkagal has quit (Client Quit)

15:13:08 [ericP]

why avoid PUT?

15:13:17 [timbl]

How does the tabr know that it should put up the "Does not exists -- Create it" button?

15:14:02 [presbrey]

~]$ curl -v http://test.rdf.me/Joe_Presbrey 2>&1 | grep -E 'HTTP/1|Content|Author'

15:14:02 [presbrey]

> GET /Joe_Presbrey HTTP/1.1

15:14:02 [presbrey]

< HTTP/1.1 404 Not Found

15:14:02 [presbrey]

< MS-Author-Via: SPARQL

15:14:02 [presbrey]

< Content-Type: text/turtle

15:14:19 [timbl]

OK

15:14:47 [presbrey]

Via: SPARQL => indicates creation is possible

15:14:56 [timbl]

a link:PotentailDocument

15:15:09 [timbl]

or DAV I assume -- why avoid DAV

15:15:23 [presbrey]

to avoid problems with concurrency

15:15:39 [presbrey]

say we are both creating the new document

15:15:50 [melvster]

melvster (~melvster@p5797F91F.dip.t-dialin.net) has joined #dig

15:15:57 [timbl]

You can't use Content-Type: text/turtle -- that is the conten type of hte 404 error message, whcih i snormally HTML

15:16:17 [presbrey]

the document I sent is turtle

15:16:31 [presbrey]

the client said they would HTTP Accept: it

15:16:54 [presbrey]

whats the problem with sending something HTTP Accept'd, regardless of response status

15:17:14 [presbrey]

if the first step of creating a new document is to PUT 0 triples

15:17:37 [presbrey]

if I start creating it and then you start "after" me, when you "start" you crush all my work when you PUT 0

15:18:06 [presbrey]

rather, if we both POST then when the round-trip refresh occurs, we see we are both creating/editing the same graph and can reconcile

15:20:28 [presbrey]

PUT would be fine if it triggered a real DAV2 locking etc., but it doesn't on dig.csail/xvm, rdf.me, anywhere!

15:23:08 [timbl]

Ok, you don't mean that field to indivate that a PUT should use

15:24:50 [timbl]

In fact BTW it isn't that weird to imagine that all documents exist and are empty. But it may be useful to distinguish nonexistent and empty later.

15:26:10 [presbrey]

ok travel safe

15:26:48 [presbrey]

for now I will suggest creation of new via curl -T, ?sparql=, and maybe ?load=

15:26:49 [ericP]

have fun in lyon, timbl

15:28:02 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

15:28:08 [ericP]

so tabulator will work even if the backend store signals non-existence with a 404?

15:28:29 [timbl]

No.

15:28:34 [timbl]

Not yet.

15:28:58 [timbl]

No, right now the store has t send an editable zero-triple document.

15:29:05 [timbl]

So Joe is proposing a change

15:29:08 [timbl]

to the protoocl

15:29:24 [presbrey]

thanks timbl!

15:30:06 [presbrey]

(but even 200 status zero-triple documents can't add new triples with tabulator)

15:30:20 [tlr]

tlr (~tlr@89.204.153.151) has joined #dig

15:30:22 [presbrey]

(not in the UI anyway)

15:30:37 [ericP]

sorry, "will" as in the future tense, not the conditional

15:30:50 [timbl]

You should be able to bey going to newdoc.rdf#foo

15:31:01 [timbl]

That should give you outlie mode with just a blue plus.

15:31:32 [timbl]

There is an existing bug I think maybe about that

15:36:33 [timbl]

?

15:36:49 [timbl]

Ohm, great .. so the snapshot will auto-update

15:36:51 [timbl]

excellent

15:37:14 [timbl]

timbl has quit (Quit: timbl)

15:42:46 [tlr]

tlr has quit (Ping timeout: 252 seconds)

16:20:16 [webr3]

presbrey, ericp - I've got two js and a php lib for full URI support with relatives etc

16:20:43 [webr3]

see: http://github.com/webr3/URI

16:22:42 [webr3]

also in: http://webr3.org/apps/play/api/rdfapi.js for js

16:48:01 [timbl]

timbl (~timbl@217.41.237.203) has joined #dig

17:41:08 [freddyt]

freddyt (~chatzilla@c-24-34-86-78.hsd1.nh.comcast.net) has joined #dig

17:42:44 [freddyt]

freddyt has left #dig

17:42:49 [freddyt]

freddyt (~chatzilla@c-24-34-86-78.hsd1.nh.comcast.net) has joined #dig

18:41:49 [timbl]

Actually presbrey has anyone checked that tabulator doesn't allow you to edit something if it has returned a MS-author-via with a 404?

18:43:08 [timbl]

timbl has quit (Quit: timbl)

19:38:27 [presbrey]

hi webr3 thanks

19:39:14 [presbrey]

rdf.me supports almost every request type and format we've ever done

19:39:56 [presbrey]

after the relative URI fixes are in place for SPARQL Update, it should be a very nice platform

19:42:58 [presbrey]

I'm going to work on better documentation and put up some examples

20:15:46 [lkagal]

lkagal has quit (Quit: lkagal)

20:20:37 [lkagal]

lkagal (~lkagal@pool-96-237-240-136.bstnma.fios.verizon.net) has joined #dig

21:02:26 [kennyluck]

kennyluck (~kennyluck@accormercurelyon0472836996.pck.nerim.net) has joined #dig

21:02:27 [kennyluck]

kennyluck has quit (Excess Flood)

21:03:21 [kennyluck]

kennyluck (~kennyluck@accormercurelyon0472836996.pck.nerim.net) has joined #dig

21:19:30 [timbl]

timbl (~timbl@93-158-30-199.subs.ibrowse.com) has joined #dig

21:38:22 [lkagal]

lkagal has quit (Quit: lkagal)

21:40:05 [jnpato_]

jnpato_ (~jnp.irc@c-98-229-160-12.hsd1.ma.comcast.net) has joined #dig

21:44:14 [jnpato_]

jnpato_ has quit (Client Quit)

21:49:31 [oshani]

oshani has quit (Quit: Mama nidi!)

22:40:45 [timbl]

timbl has quit (Quit: timbl)