- From: Bijan Parsia <bparsia@isr.umd.edu>
- Date: Wed, 12 Oct 2005 21:15:29 -0400
- To: Pat Hayes <phayes@ihmc.us>
- Cc: public-ws-desc-comments@w3.org, "Henry S. Thompson" <ht@inf.ed.ac.uk>, public-ws-desc-comments-request@w3.org, Jonathan Marsh <jmarsh@microsoft.com>, Arthur Ryman <ryman@ca.ibm.com>, David Orchard <dorchard@bea.com>
On Oct 12, 2005, at 7:21 PM, Pat Hayes wrote: >> On Oct 12, 2005, at 6:03 PM, Pat Hayes wrote: >> Arthur wrote: >>>> That is not a problem in the XPointer framework since all the >>>> parens are balanced. The WSDL URI have balanced parens. >>> >>> Sorry, I was being too brief. I realize the parens are balanced >>> within the URI itself. But consider a parser which is trying to >>> parse some notation like LISP or Common Logic Interchange Format, in >>> which the parentheses are considered to be lexical break characters, >>> and which contains embedded URIs as identifiers. Then a URI with an >>> adjacent close parenthesis on the right will be quite common, as for >>> example in a text such as >>> >>> (cl:text (ex:R ex:a)) >> >> This case, in a sense, doesn't matter since no sorta-qname convention >> I know of permits it, even with a breaking space. If the last >> character of a URI is not an ncname character, then you cannot >> abbreviate it with a qname like construct. This is why RDF/XML cannot >> serialize all legal RDF graphs. > > Sorry about the qnames. The point applies when full URIs are used, > which is correct in CLIF in any case. I just got tired of typing. My point was that plenty of legal (and common!) characters in URIs (e.g.,":", ",", "\"") are token marks in a variety of languages. But this is a side issue. It's true you cannot use the qname like abbreviation for URIs ending in funky characters for a variety of reasons (including most of the psuedo-qname syntax outright forbid it). [snip] >> URIs always must be so protected (a la NTriples) because lots of >> nasty characters (e.g., commas and quotes) can appear in a URI. > > Appearing inside is OK. Its appearing as the last character that > hurts. Can a URI have a space as its terminating character? I surely > hope not. No. > But if so, are there any URIs actually in use that do this, anywhere > on the planet? If not, then I don't really care what is technically > legal: there is a de facto standard being used. Eh. There are plenty of constructed uris that have all sorts of funky stuff in them. I'd be surprised if they were especially careful about what they terminate with. >> As I said, uris with funky trailing characters can't be abreviated >> with Qnames anyway. > > Then Im just not interested in such URIs. So I wish we weren't landed > with having to use one, for no good reason that I can determine. Either you didn't read, or you didn't understand, that there is a relevant W3C standard for constructing uri fragments, XPointer. It is extensible, not limited to XML, and reasonable. These are good reasons for using these URIs. They might not be sufficient reasons to override these other considerations, but brushing them off isn't helpful. >>> or users of these languages can be required to insert whitespace >>> before a lexical-breaking parenthesis. But all such ways introduce >>> artificiality and awkwardness into what is otherwise a very natural >>> and widely used syntactic convention. >> >> One can turn this question around and ask why we are requiring people >> to not use a composible, extensible standard (XPointer). > > Because not everyone who is using URIs is using XML, nor should they > be required to. XPointer is not XML specific, and, in this case, is not being used to identify bits of XML. One reason to resist (certain sorts of) bare names is that they *are* used (in XPointer) for identify the XML bits. There are two things to identify. The XPointer solution gives a algorithm for constructing identifiers for both. > URIs have an importance that transcends XML syntax requirements. Seems > to me that this whole discussion of XPointer is beside the point here. That's because you don't understand/have knowledge of XPointer. Here are some references: http://www.cognitiveweb.org/publications/server-side-xpointer-extreme- markup-2004.html http://www.mindswap.org/papers/swrp-iswc04.pdf We've explored using Xpointer for identifying various aspects of, e.g., OWL documents including the axioms associated with a class (as opposed to the class itself), and so forth. XPointers are extensible, composable, and have some other good features. They are, like URIs, butt ugly, for the most part. Cheers, Bijan.
Received on Thursday, 13 October 2005 01:15:57 UTC