On Tue, Jan 28, 2025 at 12:40:04AM -0500, David Birnbaum scripsit:

This leaves me still wondering whether there are rules of thumb for

choosing between using regex (e.g., analyze-string()) and using ixml

when both are available.

Have you got rules, or do you need rules?

E.g., "this string is a citation conforming to a known set of written
rules" or "I need to contract with my client that the generated text
will conform to productions of an agreed grammar". (Or "these are 80
column records with known fields".)

iXML is good for those.

If it's "the rules must be discovered", as in the sort of conversion
project were you've used fifteen passes to walk source to target in
comprehensible steps, iXML is NOT good for those. (In theory, yes,
a grammar could be written, but the cognitive load to write it is not
an especially practical undertaking.)

I am really looking forward to being able to pass those citation strings
to a grammar; having to tweeze them apart with regular expressions fails
to delight.

-- Graydon

--  
Graydon Saunders  | graydonish@fastmail.com
Þæs oferéode, ðisses swá mæg.
-- Deor  ("That passed, so may this.")