This is a public chat log generated from the #semsol IRC channel.
10:09:29
bengee: what's o['type'] =literal2?
10:09:33
what's literal2?
10:11:05
"literal2" 'literal1' """literal_long2""" \'\'\'literal_long1\'\'\'
10:11:26
that's coming from the sparql grammar
10:11:34
and thus part of the turtle parser
10:11:40
ah
10:11:44
cheers
10:12:59
a bit annoying for post-processing when a simple "literal" would have been sufficient
10:13:55
as long as i can know what all the o_types are
10:29:37
lol, bengee: http://convert.test.talis.com/?data-uri=http%3A%2F%2Fbnode.org&input=&output=html
10:30:38
you, as bnode, lots of times :)
10:31:03
heh
10:33:46
time to add an hAtom extractor, I guess
11:24:22
bengee: you booked accommodation yet?
11:24:51
yes
11:25:00
where you staying?
11:25:06
a easyhotel
11:35:20
tommorris: what should we expect semanticamp to be like? what will be expected of us, what are the demographicsgoing to be like?
11:35:38
kwijibo: It's going to be fun. :)
11:36:26
the attendee list promises a lot of fun indeed
11:37:02
tommorris: is it all swiggers?
11:37:04
In terms of expectations, we expect most people to contribute by either giving a talk or by participating in group discussion
11:37:34
No, it's not all swiggers, there are microformateers, markup mavens and members of the XML Illuminati too
11:38:05
If you don't want to give a talk, you can always get together with some other people and have a panel, or you can lead a discussion (certainty is not a requirement)
11:38:20
All the talks should be linked at least tangentially to the theme of "Web of Data"
11:39:26
tommorris: any idea what the group discussions will be yet?
11:39:38
no, that's up to you all to decide. ;)
11:39:51
in terms of the content, all I know is what I'm going to be talking about
11:40:32
which is?
11:41:15
I'm going to be giving a talk about why we should get non-technical people involved in designing metadata standards, and talking about why Strings Are Not Everything (using ID3 as an example)
11:41:31
bengee wonders if the sunday schedule is going to be created on sunday, too, or if we plan for both days on saturday already
11:41:45
yeah, what will happen?
11:41:58
What usually happens is we lay out the boards on Saturday morning and people can fill them in, but can move them during the weekend.
11:42:04
do we just mingle for an hour at the start and split into groups? or how does it work?
11:42:09
ah
11:42:12
cool, cheers
11:42:32
There is usually free space on the Sunday though so if you have a last minute idea, you can usually sneak it in later
11:42:45
The actual schedule looks something like this:
11:43:00
everyone arrives, I give a welcoming session which is about fifteen minutes long
11:43:02
bengee fears all he'll have will be very last-minute ;)
11:43:10
that's fine
11:43:31
then everyone introduces themselves - name, company/affiliation, why they are here, and three 'tags' of interest
11:43:58
then we hand out lots of slips of paper which people can use to write their name and session on
11:44:13
then they go to the area with the schedule and affix it with blue tack
11:44:24
tommorris notes to buy a few packets of blue tack
11:44:50
what kind of tags?
11:44:55
tommorris: any idea where the evening's drinking/eating will be (roughly?)
11:45:09
Oh, like "rdf" or "semanticweb" or 'rel-tag' style tags
11:45:41
cerealtom: We will either have a pub booked (not sure about that yet) or we will split up into smaller groups and go to a variety of pubs and restaurants
11:45:44
quite general then?
11:46:02
They can be specific or general, self-deprecating or serious
11:46:10
or more specific, like, inferencing, UI design, ?
11:46:13
Anything you like, just to help other people understand what your interests are
11:46:14
yep
11:46:20
tommorris: ok, thanks :) just looking for hotels so trying to decide on rough location
11:46:21
ah
11:46:49
cerealtom thinks that close to south ken is good => roll out of bed sunday morning
11:47:09
I am staying here btw - http://www.shaftesburykensington.co.uk/
11:48:02
One thing I'm hoping for is if we can get together a group of Ruby hackers interested in RDF, we might try and have a hacking session in a hotel room or something
11:48:05
Following danbri's post, the need for a decent Ruby RDF library is ever more pressing
11:49:53
has anyone run those testsuites on the native ruby libs yet?
11:50:04
i dunno
11:50:17
I tried running the 2003 test suite on the Rena library and it failed
11:50:20
is a native ruby library so important in ruby?
11:50:45
yes, I think so - for Rails apps on hosted serves, the ability to compile and use a Java or C library is not there
11:51:06
deployment of ruby is pretty awful on shared hosting anyway
11:51:17
the culture is more for shell access
11:51:25
i mean, it's not like php
11:51:31
maybe the point is, it should be
11:51:55
true, but I still think that a lot of people would probably use RDF if there were an easy, native and Rubyful way of doing it
11:52:25
The primary problem with Rena is it's literal handling is really not very good
11:52:40
would it be hard to fix?
11:53:34
I'm going to try and fix it up soon - but at the moment, I'm slightly more focused on things like badges, caterers and blu-tack
11:56:06
yeah
11:56:11
blu tack is more important
11:56:47
my hacking ambition is a jquery plugin api for traversing rdf/json
11:57:05
i'd quite like to make it like the jquery dom traversal api
11:57:11
but with an rdf path language
11:57:31
anyone had much of a look at the rdf path languages?
11:57:46
kwijibo: the place that tommorris is staying looks alright. shall i ask reception to book us both in?
11:57:50
I think I like the looks of rxpath
11:58:00
cerealtom: yeah, that'd be great cheers tom
11:58:03
cerealtom: be careful, booking online is cheaper
11:58:13
thanks tommorris!
11:58:31
kiwjibo: so you just want the one night (saturday?)
11:58:34
yeah
11:58:44
ok. any other requirements?
11:58:50
non-smoking
11:58:59
kwijibo: ok
11:59:03
cheers
11:59:13
i note that it's cheaper than that goddamned premier inn at the business park
11:59:31
kwijibo: yeah, chepaer than practically anything near BHX
11:59:46
i'm a bit sick of that place now :p
12:00:19
i think next time i'm down for any length of time, i'll try to get a b&b or something in marston green maybe
12:00:40
something more homely
12:01:32
kwijibo: yeah, can imagine. not sure whether marston green has that kind of thing, but i'm on the lookout aswell so lets share info. right now the etap in town works ok for me (but only really on financial and location criteria)
12:01:41
anyway, will mail reception
12:02:38
oh, cool, cerealtom will be there, too
12:02:56
cerealtom waves to bengee
12:03:01
Yep, the attendees list isn't totally up-to-date
12:03:02
looks like it'll be a nice swig get-together
12:03:04
cerealtom: i stayed in a b&b in marston green before the first time i was down that i booked myself
12:03:24
cerealtom smiles at the prospect of a good swig gathering
12:03:25
the white house i think it was
12:03:27
There's one or two people who I've taken off and one or two people added
12:03:59
trouble is, reception only really seem to want to book me into big motels really
12:04:43
tommorris: thanks for being accommodating :)
12:04:59
kwijibo: how about booking yourself and claiming back?
12:05:02
I don't know what you are talking about - you were on the waiting list and a place became available
12:05:46
Perfectly true. I guarantee you.
12:06:34
cerealtom: yeah, less hassle for me if reception does it, in theory, but usually there's a problem anyway, and i end up paying on my card
12:07:07
tommorris: cool, thx
12:07:23
tommorris: is there room for adw?
12:07:30
adw?
12:07:45
andrew walkingsaw
12:07:49
(I think)
12:07:54
iand mailed you about him
12:08:02
(I've reached Dunbar
12:08:04
Yes, if he can come I'll put him on the list
12:09:34
tommorris suggests continued discussion of the waiting list should occur on #semanticcamp
14:38:30
bengee: where does the parser store xmlns prefixes again?
15:35:35
$this->nsp
15:39:33
(sorry for late reply, was afk)
16:35:14
bengee wonders if "build your own sparql bot with Trice" could be a fun session for semanticcamp
16:36:33
we could perhaps try to revive foafbot, and it'd be a nice entry point to sparql and rdf for those who don't know too much about it
16:38:50
bengee++
16:39:06
I'd like to see that
16:39:55
I could show a little bit of trice that way, too
16:41:23
kwijibo, cool
16:43:40
bengee: so Trice is a Semantic Application Framework, not just a Web Application Framework ?
16:44:33
you *can* create semweb apps, but dont have to
16:45:10
i mean, it's not just for serving web documents?
16:45:29
(if you're demoing it to build an ircbot)
16:47:52
the flow is basically request -> dispatcher -> handler -> template or output or what_you_want
16:48:53
so, a handler can answer sparql requests, or resource description requests, or ordinary html doc requests
16:50:02
e.g. the ircbot has a process handler that doesn't return a result, but joins an irc channel and logs activities to a sparql store
16:50:45
another ircbot handler generates the html/hAtom logs from that store etc
16:52:48
for a blog, you could have a trackback handler etc
16:55:32
and to answer your question ;) it's not necessarily web-only, but you do control things mostly via http
16:56:47
arc-bot, you don't know much, don't ya?
17:08:15
Hola semweb compadres!
17:08:31
heya peterkz
17:08:55
better scalability is in the works, the current store is not optimized yet
17:09:11
Ah
17:09:17
you can try to tune MySQL a little bit, though
17:09:39
No problem. queries run fast but LOAD inserts get slower with triple count
17:10:16
yeah, index updates are getting expensive
17:10:58
Anyhow, I wrapped up the file system synchronisation in a plugin which is in a state of "works for me".
17:11:11
Anyone interested in testing it more?
17:11:13
the final version will have a way to split up the triple tables (and thus the indexes)
17:11:16
cool
17:11:48
Ah cool (re trple tables splitting)
17:11:56
s/trple/triples
17:12:21
bengee would love to play with the sync plugin
17:12:52
brb (diaper)
17:13:06
(for my son
17:13:14
if you dare to make it public, I could put it on the arc site
17:13:25
(the plugin, not the diaper ;)
17:36:13
Back! Here is the plugin: http://www.peterkrantz.com/2008/arc-filesync-plugin1/
17:36:54
It is a little rough but does the job for me. It is in need of better logging and documentation of course.
17:40:36
looks great
17:41:24
and you have some cron that calls it every now and then?
17:43:00
oh, he left again already
17:44:39
kwijibo subscribes to peterkz's blog
17:49:42
bengee updates http://arc.semsol.org/download/plugins
19:56:18
Sorry for dropping out earlier. Yes I run a cron job for a different task (that fetch eur-lex documents nightly). It ends with the synchronizaton call.
19:57:07
I see, cool
19:58:00
are you still adding triples, or did you stop at that 1.6M?
19:58:27
I am beginning to see that ARC will fit nicely with other stuff I am thinking of as well. I run some ruby stuff to pull down and massage data after ARC with a template layer does the presentation.
19:58:41
I stopped at 1.6M as documents took longer and longer to load.
19:58:47
'k
19:59:03
A small calculation gave that it would take about a week to load on my laptop
19:59:17
heh, not so cool
20:00:12
Well, it shouldn't be a problem really. It is just that I have a lot of docs
20:00:29
But if I can help out with optimizations I would be happy:-)
20:00:42
I owe you a lot already:-)
20:00:59
I did a test with a streaming/bulk insert on my laptop, it was down to 300 t/s at 3M triples, which was getting too slow then
20:01:18
thx :)
20:01:42
Maybe there is a difference for named graphs? Inserts were taking 6-7 secs at 1.6M triples in the store.
20:02:08
all graphs are named in arc
20:02:14
Ah sorry.
20:02:20
bt small files means index updates after each load
20:02:30
'k
20:03:29
the latest revision does an OPTIMIZE TABLEs after a number of inserts, but this is mostly for later READs
20:04:30
Ok. Select queries run really fast though. I was expecting them to be much slower.
20:04:41
that's good news
20:05:14
there is still room for improvement, but they are a lot more efficient than in ARC1 already
20:05:31
I saw you were working on a wiki on top of ARC. Another idea would be a full blown CMS of course.
20:06:13
yeah, there are plans ;)
20:06:23
http:/trice.semsol.org/
20:06:43
RDF makes sense for a lot of CMS content that is locked inside semi-proprietary table structures today.
20:06:44
Ah
20:07:21
days should be much longer...
20:07:30
:)
20:07:42
(looking at the screenshots)
20:07:45
Nice!
20:09:16
<peterkz> : RDF makes sense for a lot of CMS content that is locked inside semi-proprietary table structures today.
20:09:19
+1
20:10:04
especially for platforms like wordpress or drupal that allow third-party modules
20:10:08
Niklas (oort-guy) has been using facets to auto generate forms from model schemas.
20:10:10
Yeah
20:10:49
I would be happy to see someone developing a generic vocabulary for basic CMS stuff to enable export import between CMS products
20:10:51
bendiken is up to something for drupal, I'm really looking forward to that
20:10:59
Sounds nice?
20:11:07
s/?/!
20:11:41
I do have a vocab/namespace for the trice cms terms, but it's still evolving
20:12:19
The interesting thing about many CMS:s that allow you to create forms is that the user is basically using the CMS as a tool to create a model for their domain.
20:12:20
for session, privilege, etc
20:12:24
Ah
20:12:42
right
20:12:58
Looking forward to seeing more of trice
20:14:10
what would be really great would be multiple rdf cms thingies that could be used with either backend/data store
20:15:34
Yes!
20:16:49
I hope I'll be able to fnally launch knowee.net in a few days, after that I can focus a bit more on trice
20:18:15
and arc, and client work, and ..., ah well
20:19:01
Sometimes 24 hrs just ain't anough...
20:19:09
too true
20:20:24
no diapers here at least ;)
20:20:47
Well, gotta go and see if I can get those legal docs formatted properly. Thank you and good night!
20:20:50
:-)
20:21:08
Well diapers and their creators take time as well...
20:21:49
cya
22:15:08
regarding bengee's mention an hour or two ago, see http://flickr.com/photos/bendiken/sets/72157603487819775/detail/
22:15:14
the project itself is at http://drupal.org/project/rdf. it's not quite ready for the public eye yet, but won't be long now.