Re: Languages: The Bigger the Uglier (was: Re: Aliasing in ISO C)

|> Well we started down this path with ALGOL 60 which gave us all BNF.


> BNF is a good example why formal specifications do not save us from
> complexity, on the contrary. BNF allows us to specify complex
> grammars easily. We would create languages with simpler syntaxes if we
> had to specify it in natural language.


I can't agree. FORTRAN derived its 'syntax' in a 'natural' fashion
leading to all sorts of linguistic horrors that are still with
us. ALGOL 60 and Algol 68 have very simple syntaxes. But as with any
tool it is easy to abuse, particularly when it is merely being used as
a method of description of an existing syntax (Fortran 90 now uses BNF
to paint the broad strokes of its syntax) rather than designing a
clean syntax expressible with the *minimum* number of rules to achieve
the *maximum* expessivity. Formal specifications also allow us to
easily home in on the surprises and inconsistencies of the language -
but I suspect much use of formal language in specification ignores the
proof and corollary side.


----------------------------------------------------------------------
Dave Lloyd Email: Dave@occl-cam.demon.co.uk
Oxford and Cambridge Compilers Ltd Phone: (44) 1223 572074
55 Brampton Rd, Cambridge CB1 3HJ, UK
--

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