Re: XML Guidelines -05

[James Clark <jjc@xxxxxxxxxx>]

> 4.5 paragraph 3 is still problematic, saying "Processing instructions
> can be ignored by processors because they are not part of a document's
> character data."  On the contrary:  Section 2.6 of the XML Recommendation
> says
>
> # PIs are not part of the document's character data, but must be
> # passed through to the application.
>
> This is actually a stronger requirement than is made for elements!
> If you want to recommend that PIs not be used, you need to find a
> different rationale.

I agree with the draft's recommendation not to use PIs for normative 
protocol data structures, but I also agree with this comment that the 
rationale is weak.

I would say the reasons for not using PIs are:

1. not namespace aware; there is no way to qualify a PI target with a 
namespace

2. their use cannot be constrained by most schema languages

3. character references are not recognized within a PI

4. they don't have any XML-defined structure beyond the division between 
the target and everything else; this means that applications typically have 
to parse the content of the PI in a system-dependent way; if it was done 
with an element instead, the structure could be expressed in the XML and 
the parsing could be done by the XML parser

James