IRC log of dig on 2006-08-22
Timestamps are in UTC.
- 00:08:28 [lkagal]
- lkagal (n=lkagal1@98.68.171.66.subscriber.vzavenue.net) has joined #dig
- 05:42:09 [lkagal]
- lkagal has quit ()
- 09:27:48 [tlr]
- tlr (n=roessler@ip-83-99-50-11.dyn.luxdsl.pt.lu) has joined #dig
- 10:47:09 [Biblio]
- Biblio has quit ()
- 10:55:55 [tlr]
- tlr has quit ("bbl")
- 11:27:20 [tlr]
- tlr (n=roessler@ip-83-99-50-11.dyn.luxdsl.pt.lu) has joined #dig
- 12:08:48 [RalphS]
- RalphS (n=swick@30-7-207.wireless.csail.mit.edu) has joined #dig
- 13:10:05 [lkagal]
- lkagal (n=lkagal1@98.68.171.66.subscriber.vzavenue.net) has joined #dig
- 16:10:02 [jambo]
- jambo (i=jambo@30-6-112.wireless.csail.mit.edu) has joined #dig
- 16:55:49 [avivo]
- avivo (n=avivo@YOU.MIT.EDU) has joined #dig
- 16:56:36 [avivo]
- avivo has left #dig
- 16:58:50 [sheets]
- sheets (n=dsheets@31-33-99.wireless.csail.mit.edu) has joined #dig
- 17:44:45 [ryanlee]
- ryanlee (n=ryanlee@30-7-213.wireless.csail.mit.edu) has joined #dig
- 17:51:33 [sheets]
- timbl or DanC, opinion or fact on the difference between scheme://host/path# and scheme://host/path ?
- 17:57:30 [sandro_]
- what is that URL naming?
- 17:59:06 [DanC]
- they differ by one character; what is there to have an opinion about?
- 17:59:35 [sandro_]
- *snort*
- 18:00:41 [sheets]
- DanC, so they are different URLs or do missing fragments mean no fragment?
- 18:01:32 [DanC]
- in some cases the # makes a difference, in some cases it doesn't. But the fact that there are some cases where it does make a difference makes it clear that they're different URIs
- 18:01:56 [sheets]
- DanC, so is the Tabulator a case where it makes a difference?
- 18:02:12 [DanC]
- which part of the tabulator?
- 18:02:17 [sheets]
- DanC, display
- 18:02:35 [sheets]
- Do the two URIs refer to two different terms?
- 18:02:43 [DanC]
- display of what, the URI? yes, I would expect scheme://host/path# to look different from scheme://host/path
- 18:03:21 [sheets]
- Of course, are they different terms, though?
- 18:03:50 [sheets]
- without the fragment, I have been told is always a document if it was received with an HTTP 200
- 18:03:56 [DanC]
- I don't understand your question, I don't think. It seems trivially obvious that they're different terms.
- 18:04:25 [sandro_]
- The terms "term" and "URI" and "string" are synonyms in this context.
- 18:05:05 [sheets]
- <Person rdf:about="#"/> differs from <Person rdf:about=""/> then?
- 18:05:12 [sandro_]
- I see sheets to be asking "Does http:://foo# identify the same resource as http:://foo"
- 18:05:13 [sheets]
- I just want to be very clear on this
- 18:05:31 [sandro_]
- (sorry for the double colons)
- 18:06:10 [DanC]
- same resource: sometimes. But since sometimes not, they're different terms.
- 18:06:46 [sheets]
- Then <Person rdf:about="#"/> differs from <Person rdf:about=""/>
- 18:07:14 [sandro_]
- How can you tell, DanC? If it's HTTP does it depend on the Content-Type? With text/html are they the same?
- 18:30:43 [DanC]
- in general, the question of whether two URIs denote the same thing is arbitrarliy difficulat to determine. http:://foo# and http:://foo are two URIs.
- 18:31:31 [DanC]
- if some source you trust says, for example, { <http:://foo#> owl:sameAs <http:://foo> } , then you'll conclude they denote the same thing.
- 18:45:57 [timbl]
- Sheets, often http://foo/bar#">http://foo/bar# is used as a namespace, in other words as the start of a bunch of URIs, but not used itself to refer to anything. If it was, then the way to procxess it would be to look up http://foo/bar and find something named wit the empty string as its local name. Since neither RDF/XML nor N3 allow an empty string as a local name, that isn't going to happen in practice.
- 18:46:41 [timbl]
- There was a statement historically that an empty and missing fragid were equivalent but it was dropped.
- 18:47:57 [timbl]
- <Person rdf:about=""/> is in my opinion a false statement, as the empty URI always refs to the current *document* and documents are not people. But that is a different issue.
- 18:51:14 [timbl]
- timbl has quit ()
- 20:09:55 [ryanlee]
- ryanlee has quit ()
- 20:19:34 [RalphS]
- RalphS has left #dig
- 20:51:36 [jambo]
- jambo has quit (Read error: 110 (Connection timed out))
- 21:02:52 [sheets]
- scheme://www.w3.org and scheme://www.w3.org/ are different (one specifies a path, one does not) and yet they point to the same document and the one without a path does not redirect
- 21:11:17 [tlr]
- tlr has quit ("good night, and ....")
- 21:13:32 [DanC]
- do you mean http:://www.w3.org and http://www.w3.org/ ? Yes, those behave the same way via the HTTP protocol. They're still different URIs.
- 21:13:37 [DanC]
- I wonder what you're trying tod o.
- 21:17:28 [sheets]
- DanC, I'm tabulating those
- 21:17:41 [sheets]
- DanC, I find that they both return the same information
- 21:17:56 [sheets]
- but the metadata they return explicitly refers to the URI with a slash
- 21:19:43 [sheets]
- DanC, that is correct
- 21:20:15 [sheets]
- further confusing things is HTTP Content-Location
- 21:20:40 [sheets]
- I keep track of the content-location headers but do not smush them
- 21:21:15 [DanC]
- pronoun overload... them=? how do you smush headers? I feel like we're starting in the middle again.
- 21:23:13 [DanC]
- I don't see one.
- 21:23:29 [sheets]
- DanC, content-location is another issue that causes issues
- 21:23:39 [sheets]
- sorry for the confusion
- 21:24:06 [sheets]
- www.w3.org and www.w3.org/ have a fairly straightforward problem
- 21:24:16 [DanC]
- do tell
- 21:24:56 [sheets]
- it seems that www.w3.org should be canonicalized to www.w3.org/ as the lack of a path implies an empty path as far as HTTP is concerned (although, the URI spec seems to have changed on this)
- 21:25:26 [sheets]
- We, however, do not canonicalize URIs and so we have two URIs with effectively the same meaning that do not get smushed
- 21:25:54 [DanC]
- please, please, please, start at the beginning. I have the developer version of the tabulator on screen. http://dig.csail.mit.edu/2005/ajar/ajaw/tab.html . which URI do I start with in order to exhibit the issue?
- 21:26:19 [sheets]
- Ok
- 21:26:27 [sheets]
- start with http://www.w3.org
- 21:26:34 [sheets]
- tabulate that
- 21:26:37 [sheets]
- great
- 21:26:50 [sheets]
- now wait a little bit until the number in the upper right goes to 0
- 21:26:54 [DanC]
- roger; I see "World Wide..."
- 21:27:02 [DanC]
- 0. check.
- 21:27:15 [sheets]
- once that's done, open and close the node as the async stuff isn't working yet
- 21:27:19 [sheets]
- this refreshes it
- 21:27:37 [DanC]
- it's already open; you mean close and open?
- 21:27:43 [sheets]
- yes, sorry
- 21:27:46 [sheets]
- close and open
- 21:27:56 [DanC]
- ok, done; some yellow stuff turned green.
- 21:27:58 [sheets]
- you should see that the seeAlso links have been loaded
- 21:28:00 [sheets]
- great
- 21:28:12 [sheets]
- do you see a description of the w3c?
- 21:28:16 [DanC]
- I see an extra slash in //Overview-about
- 21:28:37 [DanC]
- no, I don't see a description.
- 21:28:40 [sheets]
- strange... you loaded the URI without a slash
- 21:28:44 [sheets]
- ok... that's the problem
- 21:28:52 [DanC]
- yes, started with http://www.w3.org
- 21:28:56 [sheets]
- there should be a description as the seeAlsos point to metadata with a description
- 21:29:05 [sheets]
- now tabulate http://www.w3.org/
- 21:29:12 [sheets]
- and you should see much of the same information
- 21:29:18 [DanC]
- do I add it to the existing state, or reload to start over?
- 21:29:21 [sheets]
- but with extra statements
- 21:29:29 [sheets]
- add it to existing state
- 21:29:42 [sheets]
- these extra statements are things like description and subject
- 21:29:50 [sheets]
- they only apply to the version with a slash
- 21:29:54 [DanC]
- yes, I see Description etc.
- 21:30:05 [sheets]
- This seems like a problem, no?
- 21:30:19 [DanC]
- no
- 21:30:25 [DanC]
- we could decide it's a problem.
- 21:30:25 [sheets]
- why not?
- 21:31:43 [DanC]
- what's your favorite flavor of ice cream?
- 21:32:05 [sheets]
- i don't like ice cream
- 21:32:08 [DanC]
- why not?
- 21:32:12 [sheets]
- too sweet
- 21:32:39 [DanC]
- ok, well, the tabulator doesn't seem like a problem cuz it's doing what I expected.
- 21:32:41 [sheets]
- ok, so why don't we decide it's a problem?
- 21:32:54 [DanC]
- I don't have strong feelings one way or another.
- 21:33:08 [DanC]
- As I have said to timbl, the whole idea of a generic RDF browser doesn't float my boat.
- 21:33:09 [sheets]
- But will most people expect www.w3.org and www.w3.org/ to mean different things?
- 21:34:16 [DanC]
- er... well, yes, based on their experience... but the semantic web deals mostly with formal systems, which often don't agree with what most people would expect.
- 21:34:26 [sheets]
- hmm... ok
- 21:34:33 [DanC]
- sorry... no, not different.
- 21:34:41 [sheets]
- even though the web server returns the same document?
- 21:36:07 [DanC]
- if you're going to try to canonicalize based on whether the server will return the same document, you'll have to look into the server's .htaccess file and such. I don't recommend that.
- 21:36:52 [DanC]
- if you want to canonicalize http://www.w3.org to http://www.w3.org/ , as a policy/design choice, that seems fairly reasonable.
- 21:37:04 [sheets]
- hrm...
- 21:37:22 [sheets]
- the next question becomes: what about http://www.w3.org/Talks and http://www.w3.org/Talks/
- 21:37:44 [sheets]
- or maybe only for hosts lacking paths at all
- 21:38:03 [DanC]
- right; hosts lacking paths is something you can reason about without going over the wire.
- 21:38:16 [sheets]
- ok
- 21:39:24 [DanC]
- in the case of /Talks vs /Talks/ , I wouldn't canonicalize them at all. They're just 2 URIs... until you go over the wire, when you'll learn they'r related. You could record that relationship as a triple.
- 21:40:26 [DanC]
- re "only for hosts lacking paths"... make that: only for http and ftp hosts lacking paths. Don't do it for arbitrary schemes. In some schemes, the difference may matter.
- 21:40:28 [evoli]
- evoli (i=yuhsin@DOMINIA.MIT.EDU) has joined #dig
- 21:40:35 [sheets]
- yes
- 21:40:37 [DanC]
- or maybe not... we should check the generic syntax spec.
- 21:40:52 [sheets]
- it's only for some schemes
- 21:40:56 [DanC]
- ok
- 21:41:00 [sheets]
- oh, i need to add https support, too
- 21:41:02 [sheets]
- bah
- 21:41:42 [DanC]
- by the way, you do realize that we're making this design up, right? there's no Right Answer to find.
- 21:42:22 [sheets]
- which design?
- 21:43:19 [DanC]
- the design for how tabulator does URI canonicalization, and how it deals with redirects
- 21:44:16 [sheets]
- ah, yes... I'd like to respect what little of the RDF, HTTP, URI, and XML specs deal with this, though
- 21:44:27 [sheets]
- it seems to sort of be at the border of all of them
- 21:44:42 [sheets]
- and there are a lot of de facto solutions out there
- 21:45:15 [DanC]
- there's nothing in those specs that says the current tabulator behaviour is wrong, is what I'm saying.
- 21:46:05 [sheets]
- right... that's what I'm finding
- 21:46:05 [DanC]
- URIEquivalence-15: When are two URI variants considered equivalent? http://www.w3.org/2001/tag/issues.html#URIEquivalence-15
- 21:46:17 [sheets]
- !
- 21:49:05 [sheets]
- ack
- 21:49:21 [sheets]
- timbl says "no definitive URI canonicalization algorithm"
- 22:37:52 [DanC]
- hmm... "I need to get in touch with alerer" says jambo re issue102. So that puts jambo back in the critical path for release. an evoli seems to be out. sheets, if you reach a stopping point, I'd like to chat for a minute about the release
- 22:38:06 [DanC]
- s/jambo back/alerer back/
- 22:38:07 [sheets]
- DanC, yes?
- 22:39:03 [DanC]
- just want to check that you're aware of these bits of news relevant to the release plan
- 22:40:18 [sheets]
- yes
- 22:40:29 [sheets]
- Not a whole lot I can do
- 22:40:47 [sheets]
- other than hound them, fix stuff myself, or remove stuff
- 22:41:25 [DanC]
- it'll be up to you to decide whether we release on Friday.
- 22:41:36 [sheets]
- ok
- 22:41:43 [sheets]
- I think it may still be possible
- 22:41:56 [sheets]
- but it may bleed for a day or two
- 22:42:09 [sheets]
- what should I do if I hit 40 hours this week?
- 22:42:09 [DanC]
- bleed... you mean slip?
- 22:42:15 [sheets]
- sure, slip
- 22:42:20 [sheets]
- but more painful :-P
- 22:43:04 [DanC]
- I suppose it can slip to the following week..
- 22:43:22 [DanC]
- anyway... just try to make changes in plans consciously.
- 22:43:45 [DanC]
- and, for example, if you decide to defer something, pls tell roundup
- 22:44:49 [DanC]
- if we slip to the next week, timbl will likely get back in the loop. That could be a feature or a bug; he might insist on getting more stuff fixed before release.
- 22:45:23 [sheets]
- roger
- 22:48:51 [DanC]
- family time in 12 minutes, FYI.
- 22:49:23 [sheets]
- no problem
- 22:49:30 [sheets]
- I'll keep chugging
- 23:52:10 [timbl]
- timbl (n=timbl@146-115-112-112.c3-0.lex-ubr1.sbo-lex.ma.cable.rcn.com) has joined #dig