- From: Gisle Aas <gisle@ActiveState.com>
- Date: 03 Feb 2004 10:03:13 -0800
- To: Julian Reschke <julian.reschke@gmx.de>
- Cc: uri@w3.org, Paul Hoffman <paul.hoffman@vpnc.org>
Julian Reschke <julian.reschke@gmx.de> writes: > I think an area where rfc1738bis *should* say something is in the area > of mapping non-ASCII (or reserved ASCII) characters in path segment. > > For instance, what are file URLs for the following Windows file names? > > 1) "c:/Dokumente und Einstellungen" > 2) "c:/a#b" > 3) "c:/ä" > > I guess everybody agrees on the answers for 1) and 2) (hex-escape the > reserved character) I don't know about this. The perl URI module does this: $ perl -MURI::file -le 'print URI::file->new("c:/a#b", "msdos")' file://c:/a%23b The reason is that this is the only way that gave sane results when resolving relative URIs according to the standard URI rules. Alternative mappings are: file:/c:/a%23b file:/c:a%23b but these give strange results of you use them as base with relative URIs like "./foo" and "../foo". abs("./foo", "file:/c:/a%23b") ==> "file:/c:/foo" abs("../foo", "file:/c:/a%23b") ==> "file:/foo" abs("./foo", "file:/c:a%23b") ==> "file:/foo" abs("../foo"), "file:/c:a%23b") ==> "file:/../foo" Regards, Gisle
Received on Tuesday, 3 February 2004 13:03:51 UTC