Re: Formal semantics of language semantics

"Dr A. N. Walker" <merlot!anw@mailbox1.ucsd.edu> writes:
|>
|> The answer is still "yes". It is "easy" to write a vW grammar
|> to emulate a universal Turing machine, and "hence" to write a vW
|> grammar that does anything whatsoever that is "interesting", ...


Gug. One then has the problem of how you specify the rules for
writing your van Wijngaarden grammar ....


I suppose that I should learn more about them :-(


|> >The aspects that I was trying to define formally were things like
|> >aliasing rules in an Algol/Fortran/C-like language, [...]
|>
|> Yes. If you can build such things into a compiler, for either
|> static or dynamic checking, you can build them into a vW grammar.


That covers most cases. There are fairly often some rules that
cannot be specified by a Turing machine program, but they are
usually less important. They usually relate either to eventual
terminal constraints or to necessarily unspecifiable conditions.


|> > And
|> >similarly with exception handling, where practical programmers are
|> >dealing with necessarily unclean exceptions.
|>
|> Build it into your byte code .... Whatever your hardware
|> can do, a vW-grammar emulation of it can do the same.


This is an example of when you can't do that. But I could specify
a set of checkable rules that would leave only a little hand waving.


|> I repeat -- the only *real* problem is that doing it generates
|> *huge* grammars that have *ginormous* productions. We're talking
|> large books to include the grammar, and large universes to include the
|> productions.


Ugh.




Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email: nmm1@cam.ac.uk
Tel.: +44 1223 334761 Fax: +44 1223 334679

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