IRC log of dig on 2007-07-04

Timestamps are in UTC.

00:58:04 [lkagal]

lkagal (n=lkagal1@79.68.171.66.subscriber.vzavenue.net) has joined #dig

02:45:36 [lkagal]

lkagal has quit ()

03:43:06 [djweitzner]

djweitzner has quit ()

10:58:16 [lkagal]

lkagal (n=lkagal1@79.68.171.66.subscriber.vzavenue.net) has joined #dig

10:58:41 [lkagal]

lkagal has quit (Client Quit)

12:33:25 [timbl]

UnicodeDecodeError: 'ascii' codec can't decode byte 0x8e in position 128: ordinal not in range(128)

12:33:40 [timbl]

Why an acii code error when utf-8 encoding?

12:55:37 [sbp]

timbl: you have an '\x8E' in the str to be encoded, i.e. before encoding

12:55:50 [sbp]

it's trying to decode that input as ascii before encoding it as utf-8...

12:56:05 [sbp]

presumably you'll need to decode it as iso-8859-1 first before encoding it

12:56:36 [sbp]

>>> '...\x8e...'.encode('utf-8')

12:56:36 [sbp]

Traceback (most recent call last):

12:56:36 [sbp]

File "<stdin>", line 1, in <module>

12:56:36 [sbp]

UnicodeDecodeError: 'ascii' codec can't decode byte 0x8e in position 3: ordinal not in range(128)

12:56:36 [sbp]

>>> '...\x8e...'.decode('iso-8859-1').encode('utf-8')

12:56:38 [sbp]

'...\xc2\x8e...'

12:56:40 [sbp]

>>>

12:56:42 [sbp]

like that

12:57:24 [timbl]

This is in teh N3 serialization stage.

12:57:37 [timbl]

So everything it is encoding shojuld be unicode.

12:58:39 [sbp]

so the \x8e input byte is actually representing the codepoint U+008E?

12:59:17 [sbp]

the input string can't be a unicode string, otherwise you wouldn't get an error

12:59:18 [sbp]

>>> u'\x8e'.encode('utf-8')

12:59:19 [sbp]

'\xc2\x8e'

12:59:19 [sbp]

>>>

12:59:23 [timbl]

I think I get it. It is encoding a string whcih it read from a file, a stdout error log

12:59:40 [timbl]

without decding it properly.

12:59:45 [sbp]

aha

13:00:21 [timbl]

that makes sense. I'll assuem all files are encodined in utf8

13:00:44 [timbl]

What I haven't foudn is the equivalent to .deecode in Javascript.

13:01:24 [timbl]

This is all testing hte n3 parser ported to js

13:17:56 [lkagal]

lkagal (n=lkagal1@79.68.171.66.subscriber.vzavenue.net) has joined #dig

13:45:42 [sbp]

timbl: the only thing I've found so far is http://developer.mozilla.org/en/docs/nsIScriptableUnicodeConverter which is a Gecko interface for doing bytes <-> unicode conversion

13:46:24 [sbp]

there's an example of using it to read a file in a user-specified encoding at http://developer.mozilla.org/en/docs/Reading_textual_data#Reading_strings_2

14:34:43 [timbl]

Pitytey call everything ns .. makes it socially more diff for others to stdize later. Noce find. Thank you!

14:35:25 [sbp]

yup. you're welcome

14:36:00 [timbl]

They have inheritance grahs in gif (which seem not to work) but no inheritnce data in rdf... ne day!

14:36:25 [sbp]

I wonder what the GIFs are backed by, if anything? UML?

14:38:12 [timbl]

or just the IDL maybe

14:42:07 [timbl]

Ok, so I will have that in the browser btu not in the rhino environment. But Rhino has access to te java runtime, so I am sure I can find an equivalent.

14:43:22 [sbp]

why the use of Rhino? a standalone Tabulator?

14:43:57 [timbl]

For testing the n3 parser in the cwm development environment - batch

14:44:19 [timbl]

http://www.w3.org/2000/10/swap/pyjs/Makefile

14:44:20 [sbp]

aha

14:44:48 [timbl]

It might even be useful for debugging

14:45:06 [timbl]

I found a bug in the store wit hit. At least it gives stack traces.

14:45:21 [timbl]

Actually has a simple debugger, though not as nice as Firebug

14:46:30 [timbl]

It is kinda weird. notation3.py was the first bit of code whcih started all of cwm etc off, now I am dealing with it again. danc emailed it t me and installed ython and played with it on a plane in 2000

14:46:35 [sbp]

there's also http://www.mozilla.org/js/spidermonkey/ - same kind of thing, but in C

14:46:45 [sbp]

might be faster. it's the implementation of Javascript used in the browser

14:47:08 [sbp]

I think you can do apt-get install smjs or something, too, on debianesque distros

14:47:17 [sbp]

heh, yeah

14:47:56 [timbl]

Rhino I had tried before and not had the enegerym but this time i foun it is just a case of sticking the tag file unwraped into classpath.

14:48:16 [sbp]

interesting that after all of the formalisation attempts, too, the original simple recursive descent module is still tops

14:49:02 [sbp]

ah. I wonder how much harder it'd be with Spidermonkey; I might give it a go and report on it myself

14:50:29 [timbl]

topos? i haven't times it.

14:50:37 [timbl]

The predictive parser should be faster

14:50:56 [timbl]

as it can jump to the right production as a function of te next character

14:51:45 [sbp]

tops in terms of being still the main parser in cwm, I mean

14:52:06 [sbp]

is that because all the other parsers simply lag in terms of features?

14:52:14 [sbp]

are they harder to maintain and upgrade?

14:58:12 [timbl]

I have never gotten around to integrating them and testing them. You made one or two, didn't you?

14:58:20 [timbl]

n3p?

14:58:33 [sbp]

yeah, n3p is the most recent

14:58:43 [timbl]

The predcitive parser I only had parsing but not generting any triples.

14:58:51 [sbp]

its test suite might be quite useful too

14:58:59 [timbl]

did you get n3p to the point where it would generate triples?

14:59:12 [sbp]

yep, I did indeed

14:59:21 [timbl]

Tets suite? ah. Yosi is accumlatibe test suites.

14:59:47 [timbl]

swap/test/n3

14:59:52 [sbp]

though with the 2004 version of n3.n3. when I tried to feed in the latest n3.n3 last year, with all the new unicode stuff, rdflib (which is used in the schema-parsing stage only) started to complain about stuff. I didn't get it fixed

15:00:18 [sbp]

yeah, I saw the HTML table of the test output you linked to in here a while ago

15:00:22 [timbl]

Ah .. the 32 bit regexps

15:00:25 [timbl]

64 bit

15:00:42 [timbl]

Oh, ok, so Yosi has it in hos test mtrix?

15:01:01 [sbp]

right, but I don't think the regexps were the problem. it was just an API change in rdflib... though of course since I didn't get past that I don't know if there'd be further problems

15:01:12 [timbl]

ah

15:01:12 [sbp]

nope, n3p isn't in the test matrix. nor its test suite, I believe

15:01:54 [timbl]

Do you have CSV wriet access to swp?

15:01:58 [sbp]

nope

15:02:11 [timbl]

I wonder whether w could get you that

15:02:52 [sbp]

for me to integrate n3p into the test suite?

15:03:10 [timbl]

and the n3p tests

15:03:25 [timbl]

we should be moving toward publishing turtle with tests too

15:03:40 [timbl]

so having a go test matrix will be important

15:04:00 [timbl]

Currently I don'r have a class of turtle tests which don't use extra n3 features

15:04:22 [sbp]

publishing turtle in TR space? which group is responsible for that?

15:05:38 [timbl]

Ivan felt that the sw interst group should do it

15:05:56 [timbl]

as a note

15:06:03 [sbp]

interesting

15:06:07 [timbl]

to get it out there

15:06:11 [sbp]

good idea

15:06:19 [timbl]

then the ietf would annoint the mime type

15:06:32 [sbp]

they only need a Note? hmm

15:07:04 [sbp]

what's the hold up of the N3 MIME type, by the way? similar situation?

16:04:13 [davidli]

davidli (i=dli@MACGREGOR-FIVE-FIFTY-NINE.MIT.EDU) has joined #dig

16:06:18 [davidli]

davidli has quit (Client Quit)

16:15:03 [timbl]

The IETF said on a liason call tehy need some 'stable document to refer to

16:15:06 [timbl]

a note would do

19:36:59 [lkagal]

lkagal has quit ()