Re: Looking for new language features

> [New syntax added on the fly? Lots of dead extensible languages from the
> 1970s like EL/1 and IMP72. -John]


I don't know anything about EL/1, so I can't comment on it
(references, especially on the web, anyone?)


I can see why you're so down on IMP. I haven't used it personally,
but it looks like every program you want to write requires a set of
productions to generate the source language you wish to work in. I
can certainly understand why you would find such a language
distasteful.


However, keeping in mind that this language is now nearly 30 years
old, and most people find 30-year old languages to be distasteful
anyway, I'm wondering if the concept of a syntax-extensible language
couldn't be brought up to date.


For my own purposes, I'm not interested in a language that forces you
to supply a grammar with every source file you supply. I'd be more
interested in a tool that lets me easily write DSELs (domain specific
embedded languages). Specfically, I'd like to have a full featured
imperative language like C or Pascal (hey, 30-year old languages!)
that could be used to write real programs "out of the box" plus the
ability to add control constructs, data types, and other syntactical
elements as appropriate for a given problem domain.


While I certainly agree with your assertion that such a feature could
be easily abused by the general programming public, I see a different
use for this rather than every source file presenting a different
source language. I see engineers creating "libraries" of language
extensions that could be used in a given problem domain. I would
expect many of these to be standardized, much like the C standard
library gave us standard string functions, math functions, etc.


Certainly, language and compiler technology has progressed sufficiently
in 30 years to allow us to do something a little more reasonable than
IMP72.
Randy Hyde
[There was a standard set of productions that everyone used to give a
common base language for IMP72, but everyone also used just enough
extra syntax to produce mutually hostile dialects. More to the point,
application libraries are swell, but we've been building them for 50
years without adding syntax to the base language. Maybe I'm just
insufficiently imaginative, but it's hard to see a lot of places where
libraries that add syntax would be worth the confusion they caused.


Most PL/I programmers treated the language as one of three slightly
overlapping subsets, the Fortran part, the Cobol part, and the Algol
and system programming part. If you were writing in Fortranish PL/I
and threw in pictures or lists, other programmers would hate you
because they had to go look up what that syntax did in order to work
on your program. I expect syntax libraries would do the same thing.


Re EL/1, it was Tom Cheatham at Harvard in the 1970s, but I can't
find any of the reports on the web. He has a home page at
http://www.deas.harvard.edu/users/faculty/Cheatham/cheatham.html,
maybe he has the old reports somewhere. -John]

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