Re: What's the word for...

tjj@netnews.summit.novell.com (CNS-ksf-+Jordan T.J.) writes:
|> > Could someone please tell me what the word is for a language
|> > which can be written in itself?


and hagerman@ece.cmu.edu (John Hagerman) replies:
|> I don't know the answer, but I'm curious: are you interested in this
|> academically or practically? The difference I mean is that while this may
|> be impossible in the "pure" form of many languages (eg, Pascal), it will
|> still (usually) be possible for real implementations of those languages...


While I think John H. is making the most important point, I'd like to add
that I believe the correct answer is "language". :-)


Every language past a certain richness can be self-referential by encoding
every symbol in the language and then manipulating the codes (that's
Goedel). Offhand, I can't think of a language with dynamic memory
allocation and/or file opening/closing that doesn't have this power.


Goedel then goes ahead and build on Russell and Whitehead to show that
some functions were logically undecidable. (Turing, of course, showed
that others were mechanically undecidable, and thus courses in
computability were born...)


Now, if T.J. wants a word for a language defined using a subset of itself
-- LISP is the canonical example? -- I hereby propose "autoliblogical",
referring to a condition where the language's libraries have now
mysteriously migrated into the compiler... :-)


BTW, semioticians probably have a word for this condition -- an English
dictionary can be written in English, after all -- but IMHO that works by
removing a necessary condition of computability: that some definitions
need to be grounded in operational (machine-operating) terms.
--
Alexander Glockner glockner@cosc.bsu.umd.edu
Asst. Professor, Dept. of Computer Science
Bowie State University Bowie MD 20715 USA
(301) 464-6609 (voice) (301) 464-7827 (fax)
--

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