"Structural pattern matching" for Python, part 1
"Structural pattern matching" for Python, part 1
Posted Aug 7, 2020 13:07 UTC (Fri) by anton (subscriber, #25547)In reply to: "Structural pattern matching" for Python, part 1 by Cyberax
Parent article: "Structural pattern matching" for Python, part 1
ML was developed by Robin Milner and accomplices in the 1973-1978 time frame. The 1978 book on Edinburgh LCF contains a detailed description of ML. It seems to me that the dominant ML dialect in this century has been OCaml.
Prolog was slightly earlier (starting in 1972); there pattern matching is a natural result of unification (unlike pattern matching in ML, which is a special language feature); the syntax, though, is not that far to what we see in ML or in the proposal. And maybe the contentious issue "Is it the value of the name or a new variable to be bound?" is there because there is no difference in Prolog (logic variables work differently from functional/imperative variables). E.g., the first clause of the definition test from the article can be written similar to the example:
test(point(X,Y),Z):- X=Y, Z="A point on the diagonal".or as:
test(point(X,X), "A point on the diagonal").Making a query for
test(point(1,2),Z) will try to unify point(1,2) with point(X,X), which fails, because X cannot be simultaneously be bound to 1 and 2.
I don't remember how the value-or-unbound issue is solved in ML.
An earlier language with pattern matching is Snobol (started 1962), but I don't know enough about Snobol to compare it.