Re: Why ML/OCaml are good for writing compilers

> From: dwight@pentasoft.com (Dwight VandenBerghe)
> [...] So although when I am writing about "ML"
> I am writing about Ocaml, what I say also applies, I think, to SML/NJ.


> So here is an unordered list of language features that seems to me
> to make writing compilers a pleasure rather than a horrendous chore.
> [...]


I heartily agree (though noting that my own experience is more with
compiler-like things than compilers themselves). And besides all
these splendid basic things there is the possibility of using
demand-driven evaluation for parsing, in caml-light with the stream
extension, and I imagine similar things are possible in Ocaml and
SML/NJ.


With caml-light streams one can write a set of mutually recursive
routines matching stream patterns against the input, often very close
to a BNF grammar notation apart from syntactic noise and semantic
restrictions,


      and galt = function
        [< gconcat e1;
              (parselist gconcat) e2
        >] -> makeseqlist (e1::e2)


      and gexpr = function
        [< galt e1;
              (parselist (function [< 'Hch `|`; galt e >] -> e)) e2;
              (parseoption (function [< 'Hch `#`; 'Hint n >] -> n)) _
        >] -> makealtlist e1 e2


with the right hand side after -> giving the semantic translation; and
the basic concepts of syntactic list and option of anything are
themselves defined in the same notation,


      let rec parselist f = function
            [< f x; s >] -> x :: (parselist f s)
        | [< >] -> [];;


      let parseoption f = function
            [< f x >] -> Some x
        | [< >] -> None;;


Another good tool in the tool-box.


--
# A E Bass Tel: (01473) 645305
# MLB 3/19, BT Laboratories e-mail: aeb@saltfarm.bt.co.uk
# Martlesham Heath, Ipswich, Suffolk, IP5 3RE
--

Return to the comp.compilers page.
Search the comp.compilers archives again.