[Bug 1330] escaping and non-ascii-characters


------- Additional Comments From mike@saxonica.com  2005-05-11 07:56 -------
Escaping restricted ASCII characters in URIs would be a disaster in practice.
Whatever the W3C specs say, we have to accommodate practical reality, and the
reality on the ground is that attributes that are supposed to hold URIs
frequently contain bits of client-side JavaScript code which will not execute
correctly on popular browsers if escaped. If you want your XSLT 2.0 processor to
be accepted by users, generating HTML that's incompatible with installed
browsers is not the way to do it.

Even escaping of non-ASCII characters causes problems in some cases, for example
in the fragment identifier of an href attribute: it creates a fragment id that's
not a valid NCName (because it contains % signs) and that doesn't match up with
the anchors in the document (which are not % escaped). Browsers do not unescape
the URI if it's used client-side. This problem is bad enough already without
making it worse.

Michael Kay (personal response)

Received on Wednesday, 11 May 2005 07:56:04 UTC