Re: Mainframe vs. Supercomputer


Nick Maclaren wrote:
<snip>
|> I found the above marginal. f66 lacked a lot.

Yes, but note what I said. Suitable languages for writing compilers in
during the 1960s and 1970s were very thin on the ground, and usually
tied into a specific platform.

But for many applications (e.g., system vendor compilers) portability
is a minor concern. This was particularly true for Fortran compilers of
that era. Vendors that maintained a variety of systems with
incompatible assembly languages often had in house languages for
compiler writing, others stuck with assembly.

Academic languages, where portability was a concern, often relied on
interpreters rather than compilers. These, in turn, were often
bootstrapped with an initial simple implementation in another language
or assembler, and then a full implementation in the language of
interest.

There weren't any portable ones much
better than Fortran 66, unless you favoured BCPL.
Even portability relying on Fortran 66 was a problem. The original
Prolog interpreter was written in Fortran 66 to provide wide
portability, but the creators complained a lot about odd differences in
behavior between systems, particularly between different IBM systems.
(I sometimes wonder if they made the common assumption that Fortran
arguments are passed by reference, or that equivalence could be
(portably) used for some of the purposes of unions.)

LISP portability
was a nightmare, C was tied up in an AT&T contract, COBOL was no better
than Fortran 66, Pascal was neither portable nor adequate, the Algols
were dying out, and what else was there? Oh, lots of essentially
system-specific ones.

What would YOU have used to write a compiler in between 1970 and 1975?
<snip>
a bootstrapping system.

.

Relevant Pages

  • Re: what is the most suitable Linux platform for Programmers and software developers
    ... C, C++, Perl, Python, Lisp, and a bunch of other languages are supported ... There are FORTRAN compilers, but FORTRAN ... Look at the Linux ...
    (comp.os.linux.misc)
  • RE: numerical libraries on integrity
    ... am sure that it should be as the Fortran or C compilers will ... not work correctly (probably, other languages). ... am sure that it should be as the linkage of .OBJs of Fortran or C ... If you have received the email in error, please notify TransGrid ...
    (comp.os.vms)
  • Re: Clearly, it is too late to fix c99 - C is dead
    ... > support pre-Standard constructs, and it is why languages that look like C ... I think the statement "Fortran has failed" is wrong. ... Already some F95 compilers have added some ...
    (comp.lang.c)
  • Re: A question on Newtons Method
    ... G95 is written in C, as are most operating systems, compilers, and ... What is OCaml written in, ... Looking at the Fortran source code from ... > archaic theoretical foundation, other languages aren't. ...
    (sci.math.num-analysis)
  • Re: Need Help: compiling old Fortran program
    ... dependent on specific compilers than Fortran). ... It is quite possible to write Fortran code that does not depend on ... I have to use only a subset of GINO for wider portability because the Windows versions are far more capable than the unix versions. ...
    (comp.lang.fortran)