Experimental IRC log swig-2009-06-03

Available formats: content-negotiated html turtle (see SIOC for the vocabulary)

Back to channel and daily index: content-negotiated html turtle

These logs are provided as an experiment in indexing discussions using IRCHub.py, Irc2RDF.hs, and SIOC.

06:36:18<gromgull>morning #swig
06:39:37<mhausenblas>morning gromgull :)
06:39:51<gromgull>:)
06:40:13<gromgull>enjoying the data mining?
06:47:43<mhausenblas>gromgull: enjoying that the network is sort of working :D
06:48:28<gromgull>yeah - only about 50% packet loss
06:53:19<mhausenblas>wh dd yu sy, gomul? ;)
06:54:43<mhausenblas>gsnedders, re your SPARQL issue, maybe wanna have a look at http://bnode.org/blog/2008/07/16/sparqlscript-semantic-mashups-made-easy
06:55:12<mhausenblas>it's implemented and AFAIK available in ARC2 (http://arc.semsol.org/)
06:55:59<mhausenblas>phenny, tell gsnedders re SPARQL, see http://chatlogs.planetrdf.com/swig/2009-06-03.html#T07-54-41
06:55:59<phenny>mhausenblas: I'll pass that on when gsnedders is around.
07:19:24<grandfatha>Hi, how can I model the following in OWL: a class A can have 3 possible datatype properties but only one can be set at any time (A hasIntValue int, A hasFloatValue float, A hasStringValue String) ?
07:20:14<grandfatha>is this the case for a allvaluesForm (unionOf ) restriction?
07:31:51<Pipian>grandfatha: Believe it's something along the lines of a restriction with maxCardinality 1 on a property that is a unionOf the other properties. But I'm not absolutely sure.
07:31:51<Pipian>(A would be a subclass of that restriction)
07:32:59<grandfatha>okay.. well I guess that would be too complicated for my code generator to map these to OO-constructs
07:33:00<grandfatha>thx
08:22:11<gromgull>ACTION wonders about smushing on data.semanticweb.org
10:23:39<tobyink>I asked on the swig mailing list, but haven't got a reply there (yet), so thought I'd ask here too. I'm looking for a predicate which means roughly "has quote" / "contains text". e.g. <http://www.w3.org/> ex:quote "Leading the web to its full potential" .
10:23:39<phenny>tobyink: 02 Jun 20:07Z <Anchakor> tell tobyink I was wondering if rather not use from sub-delims which is part of pchar, as it would be more semanticly appropriate maybe... ( http://tools.ietf.org/html/rfc3986#section-3.5 ) / or . is not much difference
10:24:14<tobyink>Anyone know if such a thing exists anywhere already? My web searches have drawn a blank.
10:24:42<tobyink>We really need a rdfs:Class/rdf:Property search engine.
10:28:41<shellac>tobyink: try falcons? http://iws.seu.edu.cn/services/falcons/conceptsearch/result.jsp?query=contains
10:30:36<BenO>tobyink, http://www.w3.org/2006/03/wn/wn20/instances/synset-quote-verb-1.rdf ?
10:31:09<BenO>tobyink, but I have to admit, I just searched in http://api.talis.com/stores/schema-cache/items and pasted in the first answer
10:33:29<tobyink>Wordnet terms can be a bit problematic. What does "wn:Person" represent - the concept of a person, or the concept of the word 'person'? What 'dc:created' date would you attach to it? The date that humans split from our nearest ancestor, or the date the word 'person' entered the English language?
10:38:58<tobyink>This http://dbpedia.org/property/quote might be right, but it doesn't have any domain/range info, so it's hard to tell. I should check with dbpedia sparql interface to see how it's used.
10:41:10<tobyink>It looks like it connects a Thing to a literal quote regarding the Thing. Whereas I want to go from an Information Resource to a literal quote taken from the Information Resource.
10:44:33<tobyink>shellac: thanks for the falcons link. Didn't find what I needed, but looks useful. I still think we can do better though with helping people find existing properties and classes. e.g. a nice feature would be to narrow down your search by domain or range. (And use RDFS inferencing to include subclasses of your specified domain / range classes.)
10:59:26<tobyink>http://open.vocab.org/terms/quote
10:59:35<tobyink>A:|ov:quote
11:00:14<tobyink>A: Usage: <http://www.w3.org/> ov:quote "Leading the Web to Its Full Potential..." .
11:28:38<gromgull>does anyone know about the internals of data.semanticweb.org?
11:28:43<gromgull>i.e. is knud here :)
11:29:13<mhausenblas>gromgull - what dya need?
11:29:17<mhausenblas>ACTION knows a bit
11:29:37<gromgull>I am writing the xml files for SFSW
11:29:49<gromgull>do I really need to specify afiliation etc. for all authors again? Surely the system should know many if not most of them?
11:29:55<gromgull>are they smushed somehow?
11:30:33<danbri>tobyink, re wordnet it depends which rdfization
11:30:53<danbri>my old 1.6 conversion (offline for now at least) was done as classes of things in the world
11:31:03<danbri>the w3c one was done as a representation of linguistic entitites
11:31:08<danbri>both are useful in their own way
11:32:16<tobyink>OK, I didn't know there were multiple Wordnet-based vocabs. (Well, I knew there were ones based on different Wordnet releases, but I thought they were both semantically similar.)
11:36:12<gromgull>mhausenblas: any idea?
11:38:22<danbri>http://lists.w3.org/Archives/Public/www-rdf-interest/1999Dec/0002.html
11:38:37<danbri>that was my original go at wordnet
11:38:58<danbri>links all crapped out of course
11:52:19<tobyink>Looks useful, so why offline?
11:52:45<libby>olden days
11:53:53<danbri>tobink, it was generated thru an unmaintainable perl script that scraped the commandline output of an old wordnet utility script
11:53:56<danbri>the w3.org copy is more stable
11:54:20<danbri>if i redid it for wordnet 3 today, i'd make rdfa pages from the sql db they have, or else tweak the script used for the w3 version
11:54:28<danbri>wordnet data is much cleaner now
11:55:32<libby>where's the w3c copy?
11:57:24<danbri>there's a Note and a namespace
11:57:29<danbri>.g w3c rdf wordnet
11:57:29<phenny>danbri: http://www.w3.org/2001/sw/BestPractices/WNET/wn-conversion.html
11:57:42<danbri>.g w3c rdf TR latest
11:57:42<phenny>danbri: http://www.w3.org/TR/rdf-sparql-query/
11:57:46<danbri>nope :)
11:58:06<danbri>http://www.w3.org/TR/wordnet-rdf/
11:59:00<danbri>ok, i need amazon ec2 help!
11:59:17<danbri>i have a running instance ... and a volume with all the data on it...
11:59:27<danbri>...i made a snapshot of the volume, and created a new volume from that
11:59:41<danbri>trying to attach that to another instance now, to ensure backups can be accessed
12:00:02<danbri>on the main server:
12:00:07<danbri>i have /dev/sdf on /mnt/foafdisk type xfs (rw)
12:00:31<danbri>but on the other, with the copy mounted, if i try: mount -t xfs /dev/sdf /mnt/foafdisk
12:00:55<danbri>mount: wrong fs type, bad option, bad superblock on /dev/sdf, missing codepage or helper program
12:00:58<danbri>and dmsg gives
12:00:58<danbri>XFS: bad magic number
12:00:59<danbri>XFS: SB validate failed
12:01:03<danbri>any ideas?
12:01:56<swh>danbri: sorry, can't help, but how are you finding EC2?
12:02:18<danbri>it seems to do the business, except when i get confused and blocked like this :)
12:02:25<BenO>danbri, all i can say is that /dev/sdf may not be a xfs mount after all -> is a ec2 snapshotted volume the same as a normal volume?
12:03:11<danbri>not sure, googling around... thanks for the pounter
12:03:13<danbri>pointer
12:03:24<danbri>how can one poke at a /dev to see what kind of a beast it might be?
12:03:39<danbri>brw-rw---- 1 root disk 8, 80 2009-05-10 16:33 /dev/sdf
12:04:01<swh>you could lookup the node numbers
12:04:08<swh>or compare to the real disk
12:05:03<danbri>(all the ebs volumes and instances are in the us-east-1a region)
12:05:05<swh>8, 80 is a SAN storage node
12:05:13<BenO>danbri, the other one is definitely /dev/sdf and not /dev/sdf1?
12:07:22<BenO>danbri, do you have cfdisk or fdisk with which to look at /dev/sdf
12:07:29<BenO>danbri, http://en.wikipedia.org/wiki/Cfdisk
12:08:12<swh>fdisk is just not right without the crazy interface :)
12:08:29<BenO>swh, heh true
12:09:51<danbri>see http://www.flickr.com/photos/danbri/3591468709/ and http://www.flickr.com/photos/danbri/3591468673/
12:10:07<danbri>(the control panel ui from amazon web interface....)
12:10:14<danbri>both seem to be /sdf
12:10:53<danbri>since i didn't roll my own .AMI, goal is to be able to reconstruct the sites from backup as quick as possible, by having databases, /etc and a dump of all debian packages ....
12:12:07<danbri>re fdisk, it's a long time since i missed with such things
12:12:13<danbri>if i run: fdisk /dev/sdf
12:12:19<danbri>Device contains neither a valid DOS partition table, nor Sun, SGI or OSF disklabel
12:12:19<danbri>Building a new DOS disklabel with disk identifier 0x5113e601.
12:12:19<danbri>Changes will remain in memory only, until you decide to write them.
12:12:20<danbri>After that, of course, the previous content won't be recoverable.
12:12:20<danbri>Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)
12:12:34<danbri>(this is on the copy ... not messing with the orig ...)
12:12:35<swh>that implies it doesn't have a partition table, so it wouldn't have a number
12:13:11<mhausenblas>gromgull: usually you'd dump from Easychair
12:13:18<mhausenblas>but I guess you don't have premium access
12:13:47<mhausenblas>anyways, I can give you the xml and you can edit by hand the XML and the run it through the conversion tool
12:14:06<mhausenblas>or you give it to me (the edited XML) and I do the rest
12:14:08<gromgull>you can give me the xml for my workshop?
12:15:16<gromgull>we don't have premium access, alas
12:15:17<BenO>danbri, hmmm... bugger. And creating copies can only be done from snapshots?
12:15:25<mhausenblas>as I said, there is none (unless you have premium access which means $$$)
12:15:25<mhausenblas>I can give you the XML from ESWC09 and you can format you input accordingly
12:15:25<mhausenblas>I plan to do the same for SPOT
12:15:27<mhausenblas>ok, look, gromgull
12:15:40<danbri>on the image that does have a mounted volume on /dev/sdf i also get "Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)"
12:15:57<gromgull>I already have the example XML though - I can just edit this for SFSW
12:16:07<gromgull>although if you ahve the ESWC09 XML maybe many of the authors overlap
12:16:19<mhausenblas>if you pay me a beer later on I'll do it for you
12:16:19<mhausenblas>ok?
12:16:19<mhausenblas>:)
12:16:19<mhausenblas>that would be great
12:16:19<mhausenblas>then hand it over
12:16:37<mhausenblas>ok. after lightning talk session
12:16:37<mhausenblas>I give you the ESWC XML dump
12:16:43<gromgull>but finding emails/affiliation for all the authors will be damn boring
12:17:17<mhausenblas>and then you may return me the SFSW dump and I take care for the rest
12:17:17<mhausenblas>(conversion, updating the dogfood server, etc.)
12:17:17<mhausenblas>yeah
12:17:18<gromgull>and for all PC members I guess
12:17:28<mhausenblas>yep
12:17:34<gromgull>right
12:17:34<mhausenblas>let's chat after the LT
12:17:39<gromgull>ok :)
12:17:50<mhausenblas>ok
12:19:24<BenO>danbri, which AMI are you using? same for both? and does the second one have xfs/xfsprogs installed?
12:20:14<danbri>same as both
12:20:21<danbri>didn't install xfsprogs
12:21:06<mhausenblas>gtg, see you all at the lightning talks ;)
12:21:21<danbri>installed them both, ... but same error still
12:22:45<swh>danbri: does the one that works have a partition table? sorry I thought that error was from the working one
12:23:02<swh>you could also try file-ing it, file might know what it is
12:23:37<swh>could be that's there's some LVM involved
12:28:31<shellac>danbri: (catching up) there's no partitioning involved. put fdisk away!
12:29:19<BenO>shellac, was just trying to see if fdisk could tell what it was
12:29:35<BenO>danbri, is there anything appearing in /var/log/messages where you try to mount it?
12:29:55<BenO>danbri, ie tail -f /var/log/messages & then try to mount?
12:37:02<danbri>yep the errors i mentioned earlier, bad magic number / SB validate failed
12:37:40<danbri>fdisk is equally confused by the happy and unhappy copies
12:37:56<danbri>the magic number / sb validate errors are from mount
12:38:07<BenO>danbri, the fdisk was just a shot in the dark :)
12:38:10<shellac>ignore fdisk. I think you want xfs_check, or something like that
12:38:18<danbri>worth a try anyway
12:38:38<BenO>danbri, xfs_check man page-> http://linux.die.net/man/8/xfs_check
12:40:13<shellac>ACTION discovers fsck.xfs does nothing
12:40:28<danbri>ok, there's a difference here
12:40:36<danbri>on the main instance with the disk mounted:
12:40:51<danbri> xfs_check /dev/sdf :
12:40:58<danbri>i get 2 lines
12:40:59<danbri>xfs_check: /dev/sdf contains a mounted and writable filesystem
12:40:59<danbri>fatal error -- couldn't initialize XFS library
12:41:31<danbri>vs
12:41:32<danbri> xfs_check /dev/sdf
12:41:32<danbri>xfs_check: unexpected XFS SB magic number 0x00000000
12:41:32<danbri>xfs_check: read failed: Invalid argument
12:41:32<danbri>xfs_check: data size check failed
12:41:33<danbri>cache_node_purge: refcount was 1, not zero (node=0x839c670)
12:41:33<danbri>xfs_check: cannot read root inode (22)
12:41:35<danbri>bad superblock magic number 0, giving up
12:42:01<BenO>danbri, there is xfs_repair that might help?
12:42:17<BenO>see if it can repair it/fix whatever has gone wrong...
12:43:55<danbri>trying. it's busy looking for 2ndary superblock...
12:44:16<danbri>logger, pointer?
12:44:16<logger>See http://chatlogs.planetrdf.com/swig/2009-06-03#T13-44-15
12:44:44<danbri>Sorry, could not find valid secondary superblock. Exiting now.
12:45:43<BenO>danbri, urgh.... thats not good news
12:46:12<danbri>well it's all trashable data, was just a clone
12:46:18<danbri>but i like it when these things are, y'know, reliable!
12:46:26<shellac>ACTION hops to #foaf
12:46:55<BenO>danbri, reliable? amazon? .....
15:07:36<danbri>EC2 problems are resolved, for those following along
15:07:52<danbri>i made a fresh snapshot of the volume, and new volume from that snapshot; which mounted perfectly
15:07:56<danbri>this time it took seconds
15:08:10<danbri>last time many minutes, i guess it timed out for some reason and corrupted itself without realising
15:09:29<BenO>danbri++ good to hear it was just amazon being flakey and not an endemic issue
15:09:52<BenO>danbri, well, I guess flakeyness is a big issue, but it is Amazon ;)
15:10:42<shellac>BenO: yeah, not sure amazon being flakey is very reassuring :-)
15:10:46<danbri>heh well it worries me a bit, but am glad the problem was prior to creating something
15:10:53<danbri>during, rather
15:11:02<danbri>i mean, it clearly never created a usable volume
15:11:15<danbri>better that than it seeming to work, then dieing at the most inconvenient time
15:11:28<BenO>danbri, shame it didn't put its hand up and say, 'Hey, I kinda screwed that one up'
15:11:30<shellac>otoh danbri and libby know a couple of people there, so maybe they can get premium support
15:11:37<danbri>btw shellac, re xfs, i have retained no memory of ever choosing xfs over another system
15:11:45<danbri>though i guess i did at some point
15:12:03<BenO>danbri, its speedy for deleting very large files ...
15:12:05<shellac>as edd said, it does use it in one of the examples
15:12:33<danbri>yeah, probably just followed a tutorial
15:12:55<danbri>if in doubt, follow the crowds cos when they get burned that'll probably be google-able too
15:13:24<shellac>worked for the lemmings, why not foaf-project?
15:30:14<gsnedders>How does the MAX() function in ARQ's implementation of SPARQL work? I was hoping I could do something like http://pastebin.ca/1446321
15:30:14<phenny>gsnedders: 06:55Z <mhausenblas> tell gsnedders re SPARQL, see http://chatlogs.planetrdf.com/swig/2009-06-03.html#T07-54-41
15:32:53<gsnedders>hhalpin: hey. When are you back in Edinburgh?
15:36:26<LeeF__>gsnedders, i think it's more like SELECT MAX(?date) ... dc:date ?date
15:36:34<LeeF__>i.e., the function is selected out, rather than put in the triple pattern
15:38:29<tobyink>I think it's as per LeeF__'s suggestion - i.e. analogous to how aggregate functions work in SQL.
15:40:00<tobyink>That makes it pretty difficult to do what gsnedders wants without subqueries (which haven't been standardised in SPARQL yet).
15:40:49<danbri>lemming is now stuck with testing backups
15:40:58<danbri>hey everone, testing your backups is great!
15:41:26<danbri>seems "mysqldump --all-databases " isn't dumping enough to recreate mysql users and their passwords
15:42:28<mattl>danbri: what died?
15:43:01<danbri>nothing :)
15:43:08<danbri>i'm being proactive or something
15:43:33<danbri>writing down what you need to do to go from the backups of www.foaf-project.org to making a running ec2 instance
15:44:56<mattl>danbri: happy to mirror foaf-project.org
15:45:29<csarven>tobyink What was the URI that you had transformed http://identi.ca/notice/4621236 into RDF?
15:45:43<tobyink>Being proactive as a reaction to this though - <http://esw.w3.org/topic/FoafDowntime>. Reactively proactive - is that an oxymoron?
15:46:24<tobyink>csarven: http://buzzword.org.uk/2009/microturtle/implementation.cgi?acct=tobyink&format=text/turtle&notice=4621236
15:46:28<danbri>mattl, having offsite backups would be great!
15:46:51<csarven>tobyink Thanks
15:47:06<danbri>tobyink, well it's an improvement on the old setup, which was backed up but hard to restore quickly, esp as we didn't realise we'd been hacked for a while
15:47:48<tobyink>Which reminds me - I didn't have an opportunity to run my backups last weekend. Will have to do tonight.
16:15:03<danbri>ok
16:15:41<danbri>mysql> flush privileges;
16:15:43<danbri>I HATE THAT
16:15:47<danbri>you wrack your brains, and then it is so simple
16:18:34<PovAddict>lol
16:18:48<PovAddict>get postgres :)
16:19:33<tobyink>I've certainly always found postgresql's command line client a lot easier to use than mysql's.
16:22:11<tobyink>tab-completion is a big win; inline help for SQL syntax too.
16:24:21<tobyink>Interesting syntax idea for microturtle - when you use a CURIE the prefix becomes the default.
16:24:54<tobyink>e.g. [ a foaf:Person ; :name "Toby" ] :made [ a :Document ; dc:title "ttldent" ] .
16:25:53<tobyink>:name, :made and :Document automatically take "foaf" as a prefix because they follow the initial foaf:Person.
16:25:59<tobyink>Makes sense?
16:26:40<PovAddict>first time I used postgres i was confused by all the extra stuff
16:26:56<danbri>hey, what's the best scripty way to recreate a set of users from one unix box to another one? preserving whatever IDs stop file system getting confused,etc?
16:27:09<danbri>won't be a huge number of users anyway but still
16:27:19<PovAddict>danbri: copy /etc/passwd? :)
16:27:47<danbri>something in me screams "no..." there
16:27:51<tobyink>Copying /etc/password is only a first step - you normally need to create homedirs too.
16:28:02<danbri>though should all be ssh pubkey logins
16:28:11<danbri>and homedirs can be in the mounted volume
16:28:26<PovAddict>danbri: /etc/passwd doesn't have passwords :P
16:28:32<tobyink>Not these days, no.
16:28:34<PovAddict>passwords are stored in a different file nowadays
16:28:50<danbri>when there's other state somewhere, wonder about debian getting vewwy confused
16:29:15<PovAddict>first time I used postgres i was confused by all the extra stuff, schemas, catalogs, tablespaces... someone told me all good RDBMs have that, and MySQL is a toy engine for not having them :P
16:29:19<tobyink>The /etc/shadow file has passwords, if they're on the filesystem at all.
16:30:35<tobyink>I needed to keep a few boxes synced like this once, and just wrote a quick perl script that copied all entries with a UID > 500 from one /etc/passwd file to another.
16:30:49<tobyink>UIDs < 500 tend to be used by systemy things.
16:31:11<tobyink>So I left it to the package manager, etc to deal with them.
16:31:35<PovAddict>if the systemy users aren't there already, they are created when the relevant package is installed
16:32:39<swh>you need /etc/shadow on modern machines too
16:33:03<PovAddict>yeah if he wants to copy passwords
16:33:17<PovAddict><danbri> though should all be ssh pubkey logins
16:35:43<danbri>I reckon .... grep "x:1[0-9][0-9][0-9]" /etc/passwd
16:36:00<danbri>...then run adduser with each, specifying --uid and the homedir path
16:36:28<danbri>i mean, grep from the archived system
16:36:36<swh>sounds like a job for awk | sh
16:36:44<swh>or sed | sh
16:37:21<danbri>i'm scared of over-automating this, to be honest! i'd rather have a 2 page readme
16:37:27<danbri>it isn't supposed to happen often
16:37:44<PovAddict>type the user names and IDs by hand then? :)
16:39:39<danbri>yeah, pretty much
16:40:01<danbri>btw any rdflib OSX users know how to suppress this? /Library/Python/2.5/site-packages/rdflib-2.4.0-py2.5-macosx-10.5-i386.egg/rdflib/sparql/bison/SPARQLEvaluate.py:196: DeprecationWarning: There is the possibility of __repr__ being deprecated in python3K ... combinationLambda = 'lambda(i): %s'%(' or '.join(['%s'%mapToOperator(expr,prolog,combinationArg='i') for expr in reducedFilter]))
16:40:08<danbri>or fix it, even? :)
16:41:37<PovAddict>ARGH any idea how i can check ehat stupid app is grabbing the keyboard on X?
16:42:15<PovAddict>oh
16:42:17<PovAddict>g'damn it
16:42:39<PovAddict>ACTION needs to see what's up with this keyboard cable -.-
16:42:52<PovAddict>ACTION is still glad he can send emergency IRC messages from his ipod
16:44:36<gsnedders>LeeF, tobyink: Thanks
19:28:18<kwijibo>is there a datatype URI for fahrenheit ?
19:37:10<KjetilK>hmpf
19:37:19<KjetilK>ACTION hands kwijibo the SI system ;-)
19:37:45<kwijibo>.g SI system temperature
19:37:46<phenny>kwijibo: http://en.wikipedia.org/wiki/SI
19:38:00<KjetilK>kwijibo, kelvin
19:38:02<KjetilK>:-
19:38:07<KjetilK>errr, :-)
19:38:43<kwijibo>ACTION is amenable, but hands KjetilK yahoo weather rss :)
19:38:51<KjetilK>:-)
19:39:56<KjetilK>BTW, did you have any progress on the summitpost stuff?
19:40:19<kwijibo>i've got a scraper of sorts together for the routes pages
19:41:02<KjetilK>ok, cool
19:41:14<kwijibo>i'm trying get weather forecasts into RDF now so I can plug it into climbing areas :)
19:41:36<kwijibo>unfortunately I don't really know much about the domain
19:41:36<KjetilK>ah, then yr.no is a really good resource
19:41:51<KjetilK>actually, I'm hoping to get time to get them to do it
19:42:19<KjetilK>you get XML dumps from there, but they are not really very suitable in their current form as Linked Data
19:42:32<kwijibo>that would be cool :)
19:42:36<KjetilK>very
19:43:31<KjetilK>BTW, I got the names and nicks of the key people at summitpost, but not their emails
19:43:45<KjetilK>we could send them a PM on the site though
19:43:49<kwijibo>where do they live on irc ? :)
19:43:57<KjetilK>hehe
19:44:19<KjetilK>summitpost nicks, I guess... :-)
19:44:55<KjetilK>summitpost should be linked to geonames
19:46:06<KjetilK>then, I'm hoping they will link yr.no to geonames, problem solved :-)
19:46:43<kwijibo>would you need to query geonames for the right uri for each place ?
19:47:30<KjetilK>hmmm, I don't think so, yr.no allready uses geonames for the coordinates of the world
19:48:02<kwijibo>i would form the summitpost side though?
19:48:05<KjetilK>it is just for Norway it sucks, since it uses proprietary data for Norway
19:48:07<kwijibo>*form the
19:48:20<kwijibo>*from the
19:48:21<KjetilK>yeah, right now, I guess
19:48:27<kwijibo>i was thinking of using http://placetime.com/geopoint URIs for places I have lat and long for
19:49:05<kwijibo>but not sure if that really solves it
19:49:10<KjetilK>but I guess we might be able to persuade the summitpost guys to create a UI for people to add their mountains to Geonames with a URI
19:49:27<kwijibo>because the lat and long would need to be exactly the same for them to match
19:49:51<KjetilK>yup
19:49:57<kwijibo>i mean the lat/long from the forecast would need to be the same as the lat/long from summitpost
19:50:59<KjetilK>yeah, but if you could somehow map the summitpost URIs to geonames URIs it would solve the problem too, no need for lat/long
19:51:17<kwijibo>yeah
19:52:05<kwijibo>maybe i could put geonames into a store, and add some geonames triples to the climb location data in a second stage
19:52:14<kwijibo>in batches
19:52:37<kwijibo>well, I'm sure it's solvable somehow in a not too inefficient way :)
19:52:45<KjetilK>hehe
19:53:01<KjetilK>well, you could use the geonames ontology to only ask for mountains
19:53:40<KjetilK>and map the summitpost mountain coordinates to a nearby geonames mountain with a similar name
19:53:58<KjetilK>problem is that the geonames coordinates are often pretty bad...
19:54:18<kwijibo>what do you think of this as a publish pattern for forecasts ?: :Place weather:forecast <http://weather.dataincubator.org/forecasts/now/Snowdon> .
19:54:32<kwijibo>then
19:55:05<kwijibo><http://weather.dataincubator.org/forecasts/now/Snowdon> redirects to <http://weather.dataincubator.org/forecasts/2009-06-03/Snowdon.rdf>
19:55:06<KjetilK>it would probably be a good thing to have summitpost mountain maintainer link directly to geonames, it would mean an explicit link and geonames could inherit the coordinates
19:55:20<KjetilK>yup, that'd be nice!
19:55:28<KjetilK>very webarchy too
19:55:42<kwijibo>then would there need to be some link back to <http://weather.dataincubator.org/forecasts/now/Snowdon> from that document ?
19:56:24<KjetilK>I didn't quite understand that question
19:56:30<kwijibo>because when you deref a URI, you look for it as a subject in the document you get back
19:56:43<kwijibo>oops, train, gotta go
19:56:50<KjetilK>hehe
19:57:10<KjetilK>ACTION hopes that doesn't mean kwijibo was run over by one...
19:57:32<mischat>http://sameas.org/
19:58:13<mischat>D: | "SameAs" interlinking the Web of Data
19:58:42<PovAddict>KjetilK: * kwijibo has quit IRC (Connection reset by train)
19:58:53<kwijibo>:)
19:59:08<PovAddict>kwijibo: you missed this
19:59:08<PovAddict>[16:57] * KjetilK hopes that doesn't mean kwijibo was run over by one...
20:00:03<kwijibo>:)
20:00:42<KjetilK>:-)
20:00:43<kwijibo>yeah, gotta go onto one, not out from under one
20:01:25<kwijibo>KjetilK: what I mean is, I want to have URIs like <http://weather.dataincubator.org/forecasts/now/Snowdon> to point to with weather:forecast which will always redirect to the latest forecast
20:01:40<KjetilK>yeah, I guess it is a good idea that the URI you dereference is in the document, but I haven't thought to carefully about that
20:02:20<kwijibo>so i wonder what the relationship is between <http://weather.dataincubator.org/forecasts/now/Snowdon> and <http://weather.dataincubator.org/forecasts/2009-06-03/Snowdon>
20:03:05<kwijibo>my first thought was something like dct:latestVersion - except it isn't a version, it's a different forecast
20:06:59<KjetilK>right, I don't know, any other takers?
20:10:26<kwijibo_>d/c by train again :p
20:11:09<PovAddict>train ran over the cat5?
20:12:12<kwijibo_>anyone know anything about weather ? what is 'rising' ?
20:12:24<kwijibo_>(http://weather.yahooapis.com/forecastrss?p=UKXX0896&u=c)
20:13:54<MacTed>kwijibo - "rising" would usually refer to "barometric pressure"
20:14:42<kwijibo_>ah, right, of course :) thanks MacTed
20:15:33<kwijibo_>the weather feed has both pressure and rising - how do they relate ?
20:17:21<MacTed>hrm. without yet looking at the data, pressure is probably a number, while rising ... hm.
20:17:53<kwijibo_>i dunno - both have "0"
20:22:07<MacTed>ah! if you look at the "full report" linked from the feed, I think you'll figure it out
20:22:41<MacTed>pressure is absolute current measure (in millibars); rising is falling/steady/rising (presumably -1/0/+1)
20:22:53<MacTed>see http://weather.yahoo.com/forecast/UKXX0896_c.html
20:25:19<kwijibo_>ah! :)
20:26:40<MacTed><-- repository of occasionally useful trivia ;-)
20:27:50<kwijibo_>hmmm MacTed any idea how they calculate 1016mb form that data ?
20:27:52<kwijibo_>*from
20:28:39<kwijibo_>I don't get it
20:28:44<kwijibo_>it has <yweather:units temperature="C" distance="km" pressure="mb" speed="kph"/>
20:29:11<kwijibo_>and <yweather:atmosphere humidity="67" visibility="9.99" pressure="0" rising="0" />
20:29:26<kwijibo_>so i would presume pressure is 0 mb ?
20:29:37<MacTed>I think they're not transforming properly, so the feed doesn't actually carry everything the website displays
20:29:38<kwijibo_>but in the html page, they say 1016mb
20:29:53<kwijibo_>hmm, quite possible
20:30:27<kwijibo_>weather.yahoo.com-- :p
20:31:57<MacTed>also, I think many (most? all?) people getting the feed won't notice -- because that data isn't in an HTML rendered segment
20:33:33<kwijibo_>also quite possible
20:34:54<kwijibo_>i don't see anywhere to give them feedback ..
20:43:18<kwijibo_>ACTION off for a bit to find 'lectricity and shelter
20:44:25<PovAddict>rising might mean temperature too
20:44:27<PovAddict>I guess
20:44:32<MacTed>yeah... I bet yahoo would handoff complaints to Weather.com, so you might just look there
20:44:52<MacTed>PovAddict - given that it's in the atmosphere segment, I'm betting against temperature
20:45:02<PovAddict>ah well I didn't look at context :]

Back to channel and daily index: content-negotiated html turtle