Subject: Re: More XSL Discussion From: "Michael Kay" <M.H.Kay@xxxxxxxxxxxxx> Date: Tue, 24 Feb 1998 12:28:55 -0000 |
Jeremie Miller: >How does anyone feel about some of the other things I've proposed? I don't feel you've got all the answers right, but I do think you are battling with some very real problems in the current spec. In my view the use of pseudo-HTML style tags in the action part of XSL rules is a mistake, for several reasons: - it imposes a requirement that the start and end tags be balanced. This is extremely restrictive: for example if you have two patterns, one for the first element of a group and the other for the last element in a group, you cannot generate a opening <UL> tag in the first and a closing </UL> in the second. You can solve that with CDATA, or it could be solved by a new XSL pattern that matches the whole group rather than the first or last element, but unless you greatly extend the pattern matching capability (e.g. to detect a group consisting of a <HEAD> element followed by a sequence of <PARA> and <FIGURE> elements) it will always be restrictive. - it means that the DTD for XSL is closely tied to the DTD for HTML, and since the latter is full of vendor extensions, the former will be as well. - it creates a superficial impression of a single seamless language, but the impression is only skin-deep: there are actually two separate languages, the core XSL rule/pattern/action language and the embedded HTML rendition language. I think you would get better usability, extensibility, etc, if the difference between the two were explicit and obvious rather than being papered over. Mike Kay, ICL XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
Re: More XSL Discussion, Paul Prescod | Thread | Re: More XSL Discussion, Paul Prescod |
More XSL Discussion, Jeremie Miller | Date | Re: More XSL Discussion, Paul Prescod |
Month |