ted@imsvax.UUCP (Ted Holden) (11/21/85)
The Dec 24 issue of PC rates three PC class Fortran compilers, and picks
the Lahey F77L compiler as its favorite. This compiler is fast and has all of
the extensions beyond the 77 standard which you normally get with mainframe
Fortrans. Beyond what PC has to say about F77L, I have heard several claims
to the effect that the SPSS people funded the development of this compiler
because the RM and Microsoft compilers were deemed inadequate for the develop-
ment of SPSS-PC.
If you look at mainframe Fortran manuals, half of what you'll see is stuff
which goes over the 77 standard, and that breaks two ways: about half are
features such as having characters in real and integer arrays which keep your
35 year base of software compilable, and the other half, things such as name-
list reads etc. are things which are identical from an Amdahl to a CDC to a
1184 etc. etc., but one manufacturer couldn't do it so it got left out of the
standard. On top of all of this, any reasonable system of debugging features
goes beyond the standard. F77L seems to have all of these things; previous
PC fortrans are a sorry lot by comparison. Most have ammounted to literal
implementations of the 77 standard, which plainly doesn't get it, especially
for anyone thinking in terms of transporting applications from a mainframe
to PCs or ATs. One last point of interest for people in this boat: the new
version of F77L allows arrays to go over the 64K limit.