Re: [xsl] Re: If XSLT is declarative, why doesn't it feel that way?

["Liam R. E. Quin liam@xxxxxxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

On Mon, 20 Apr 2026 12:48:32 -0000
"Martynas JuseviD
ius martynas@xxxxxxxxxxxxx"
<xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx> wrote:

> In other words, although I cannot formally prove it, XSLT offers
> fewer ways to manipulate XML data than a general-purpose language
> such as Java -- and that is a good thing :)
>

Formally, XSLT is turing-complete, as is e.g. Java, and the set of
computations they can perform (ignoring e.g. memory limits) is
identical, i think.

In the XSLT courses i run, we do some timings. People are often
surprised to see that e.g. Saxon gets 3 minutes of CPU time in 1 minute
of elapsed time - Saxon has seen an opportunity to use multiple threads
and optimized automatically.

I donbt know what about XSLT does not feel declarative. For example,
you canbt assign to variables, we donbt have looping (no, for-each is
not a loop in the procedural sense), and a named template is equivalent
in many ways to a function, but not to a procedure. It canbt change its
argumentsb values, it canbt change global variables, and calling the
same template or function with the same arguments will under most
circumstances return the same result - the processor can even cache the
result and not call the function more than once.

Strictly speaking functional, non-procedural, declarative, are not
synonyms, but there is a lot of overlap, and XSLT sits in the middle of
these.

liam


--
Liam Quin: Delightful Computing - Training and Consultancy in
XSLT / XML Markup / Typography / CSS / Accessibility / and more...
Outreach for the GNU Image Manipulation Program
Vintage art digital files - fromoldbooks.org