Jean Abou Samra <[email protected]> writes:
> convert-ly only does text substitutions based on patterns that are
> likely to be old syntax that needs to be converted. It cannot
> understand the logic behind Scheme code. In this snippet, it can't
> determine that what is being looked up in the alist is an
> `articulation-type` value. (And it's not a lack of programming
> resources but a theoretical impossibility,
> cf.
> [https://en.wikipedia.org/wiki/Rice%27s_theorem](https://en.wikipedia.org/wiki/Rice%27s_theorem).)
> Therefore, it is to be expected that Scheme code can require manual
> conversion.
convert-ly uses heuristics. There is no point in talking about
"theoretical impossibilities" because convert-ly does not claim to be
doing a 100% job.
But that's no excuse for not trying to do a 90% job instead of a 0% job.
Patterns like
("flageolet" .
(acons "flageolet"
(assoc "flageolet"
have some chance at being candidates for conversion. How bad is the
chance for false positives? Considering all articulation types,
probably non-trivial. And false positives probably weigh heavier than
false negatives concerning conversion. So it may make sense to look for
those patterns only in expressions that also talk about
articulation-type or its associated data structures.
--
David Kastrup
