Re: [xsl] Using xsl:output in browsers, was: Re [xsl] XHTML html validation

Subject: Re: [xsl] Using xsl:output in browsers, was: Re [xsl] XHTML html validation
From: Abel Braaksma <>
Date: Tue, 20 Feb 2007 14:25:15 +0100
Manfred Staudinger wrote:

Both run the MSXML libraries. Whether you call it ActiveX or differently, they are invoked and they are the same. There are several ways to find out what IE actually uses. In this age-old article (which on a side note advices against PI for a very different reason) it is explained how: Only as of IE7, this method of invocation has changed.

not popups about ActiveX, but instead, maybe, popups about cross-frame
No popup in case of iframe.

I meant this: and this:

You just have to be careful and know thy security restrictions (which are tightened in IE7 and FF as compared to IE6). This is even more so when you use the result of a transform, which behaves a little different in the security contexts (but this is also true for javascript invoked transformations).

You should consider PIs not instead, but in addition to the javascript
invoked transformation. This way you will gain additional flexibility!

I agree that you can do both, and in certain situations it may be to an advantage, but it depends of course.

You seem not to have explored this path up to now, as many of your arguments
rely on general considerations, not on concrete experience.

This is MS say on this: "Although the use of processing instructions is sometimes discouraged, their loose structure and the flexibility of their placement make them useful for sending messages to an application without disrupting the flow of XML information." (but in the same article they claim to follow the standard, which they - of course - don't:

Well, I rather not go too deep into any yes/no discussion (sorry Tommy), but I have gone this path too often and it wasn't for me for a plethora of reasons. There are ample situations where PI can be handy but it gives me the same feeling as GOTO gave me 20 years ago: it works, but just isn't "it". A PI is very simple to implement (if you don't care for hard-linking, setting your web server to non-standard content-types and content negotiation and changing the PI to non-standard values to get it working, see, but the drawbacks and the payoffs are not worth it for me. Every now and then I try it again, but it just frustrates me for the lack of possibilities and robustness. But hey, that's only an opinion of course.

Perhaps it is worthwhile to create a little list with pros and cons for people that are new to the subject so that anyone can decide for his/her own situation. I understand that you know pretty much of this type of transformation invocation, maybe you can help improving a Wiki article on this subject (not yet there).

Abel Braaksma

Current Thread