🚀 go-pugleaf

In <1160@eta.unix.ETA.COM>, skappel@eta.unix.ETA.COM (Steve kappel) asks
about third-party M2 compilers for Sun and Apollo systems.  The January
'89 Unix Review has an article comparing compilers from ana-systems
(415-341-1768) and Oregon Software (503-245-2202) running on a Sun-3.
The ana-systems compiler is also available on the Apollo and a bunch of
other 680x0 machines; the Oregon compiler is not (yet?) on other 68Ks,
but _is_ available on VAX/VMS and SCO XENIX/386.  Go figure it.

As a quick summary for anyone who cannot find the magazine, Oregon gets an
overall grade of B in the review's "Report Card" box, while ana gets a C.
The major difference in the component scores is performance, B+ vs C;
Oregon compiles faster, while generating code which is both faster and
smaller.  Both products get C- for documentation and B for support.

BTW, the article says that the Oregon compiler supports 4-byte INTEGERs
and 8-byte LONGINTs.  Can anyone confirm or correct that?  (If it's true,
I presume that there are also 2-byte SHORTINTs, so that you can map certain
Unix data structures correctly, nicht wahr?)  If true, it makes the Oregon
compiler a better fit for commercial (accounting) applications than most
implementations of languages beloved by COBOL-haters are.  I'd rather use
a 64-bit LONGINT to keep track of the national debt in pennies than have
to resort to a library-supplied DECIMAL type, where even addition and
comparison require procedure calls. :-(

Noting that the Oregon compiler runs on several architectures already,
if I were in the market for a RISC-powered workstation I would beg
Oregon for a hint as to whether they plan on targeting SPARC or MIPS
next.  (What's that you say?  88000?  <grin>)

DISCLAIMER: I haven't tried either of these compilers myself (I haven't
the iron), and I have a friend at Oregon SW, so please read the review
instead of taking my word for it.

Christopher T. Jewell   chrisj@cup.portal.com   sun!cup.portal.com!chrisj