Re: [xsl] Safe-guarding codepoints-to-string() from wrong input

[Abel Braaksma <abel.online@xxxxxxxxx>]

Florent Georges wrote:
>   The intent of XSLT transformations is not to work with
> characters, but with the data model defined in [XDM].  Of
> course, there are a lot of functions to operate on strings,
> but dealing with the codepoints directly in the stylesheet
> is not a frequent thing IMHO.
>
>   The data model and its lexical representation are two
> different beasts.
>   


I understand that. But with the advent of unparsed-text() and related 
functions, plus removing the necessity of having an input document, XSLT 
opened up a whole new world of applications of data transformations. As 
a matter of fact, I find myself working daily doing transformations of 
legacy text-based data formats.

But I'm aware of the discussions that have been in the past about XML 
compliant data and if XSLT should be able to treat non-XML characters 
(through unparsed-text() or otherwise). For this version, partly for 
security reasons (buffer overflow etc), it was decided against 
facilities for non-XML compliant data. Unfortunately. Perhaps in a next 
version (in another decade?).

Cheers,
-- Abel