xml vs json

Let's assume you're agnostic regarding input and output formats, so you're
ready to work either with xml or with logically equivalent json input and
output data.

Then we have a question: how xslt processing will compare for two logically
equivalent pipelines: where one deals with xml, and other with json?

We have several hypotheses that hint on advantage of json, like:

- json is lighter than xml to serialize and deserialize;
- json stored as map(*), array(*) and other item() are lighter than node() at
runtime, in particular subtree copy has zero cost in json;
- templates with match patterns to some extent can be efficiently implemented
for maps using lookups of functions;
To prove anything we need to commit an experiment (we're going to use Saxon as
engine).

So, our question to the community: is there an isolated small representative
xslt around xml (along with xml files) for us to use as a model to build
equivalent xslt around json?

Thanks
--
Vladimir Nesterovsky