- From: Andrew Fedoniouk <news@terrainformatica.com>
- Date: Tue, 7 Sep 2004 13:13:12 -0700
- To: "L. David Baron" <dbaron@dbaron.org>, <www-style@w3.org>
Thanks, David, I've missed character class/range (a-z) declaration. But as far as I remember such characters as '^' or '`' are not allowed in canonic URLs (rfc2396). I mean [*-~] is too optimistic. No? >If it's implemented interoperably, I don't see a good reason to change >it at this point. Is it not implemented interoperably? Didn't test yet. Andrew Fedoniouk. Terra Informatica Software, Inc. ----- Original Message ----- From: "L. David Baron" <dbaron@dbaron.org> On Tuesday 2004-09-07 11:48 -0700, Andrew Fedoniouk wrote: > Analyzing CSS2.1 and CSS3 ([1],[2]) I've found that rules > for url defintion does not include '/' character. > > So, following given rules: > > url ([!#$%&*-~]|{nonascii}|{escape})* > "url("{w}{url}{w}")" {return URI;} , > > url(images/background.png) > > cannot be treated as a valid URL. May it is better to define first rule as > url ([!#$%&*-~/]|{nonascii}|{escape})* > ? '*' is U+002A, '~' is U+007E, and '/' is U+002F, which is in the range [*-~]. So your assertion that it's not a valid URL is incorrect. (And if you're not treating '-' within character sets as a character range operator, then letters aren't valid either.) > And what about whitespaces? Following rules above: > > url(my background.png) > > is also not valid as it contains whitespace in the middle. True. > Is this limitation make sence? If it's implemented interoperably, I don't see a good reason to change it at this point. Is it not implemented interoperably? -David -- L. David Baron <URL: http://dbaron.org/ >
Received on Tuesday, 7 September 2004 20:13:57 UTC