- From: Martin Duerst <duerst@w3.org>
- Date: Wed, 10 Jul 2002 00:07:14 +0900
- To: Al Gilman <asgilman@iamdigex.net>, uri@w3.org
Hello Al, I fully agree with your proposal below. Please note that your 'best practice' rules apply not only to file:, but also to http: if you want to be sure that you can move from one server to another. Regards, Martin. At 11:21 02/07/04 -0400, Al Gilman wrote: >An interesting issue has come up in the context of the standard digital >talking book ><http://www.niso.org/standards/resources/z3986-2002.html>. [See also ><http://www.loc.gov/nls/niso/>.] > >The dominant mode of distribution for these information objects is as >file-sets >in media, for example CD ROM. > >The specification makes heavy use of URI-references from and to XML and >specifically SMIL in holding the composite object together. So references >by URI-reference are a critical dependency of this technology. > >But from the CD-ROM the files are referenced by file: URLs. > >In the context of file: URLs it would appear that there is no common practice >as to the case sensitivity of file paths. So we came up with the following >candidate rules for file naming and reference within a talking book published >as a fileset to be safe under conditions of media distribution and file-system >access: > >1) the file paths to distinct information objects (within the book) should >be distinct under case-insensitive comparison. > >2) references across files within the book should match under case-sensitive >comparison. > >This pair of constraints-on-practice should make the set of file-paths >within the book a namespace, and the URI-references a graph encoding, >which are insensitive to the case-sensitivity characteristics of any >given local file system. > >Is this an established FAQ in any community of which you are aware? > >Is there any good reason not to impose these constraints on practice in this >application? > >Al
Received on Tuesday, 9 July 2002 11:07:42 UTC