[comp.sys.dec] Fortran/Latex summary

chinch@theory.tn.cornell.edu (10/19/90)

A few days ago I had inquired about Fortran/Latex on the DecStations.
I received only 1 mail message, and here it is :


> 
> I had a few questions about the Decstation 2100 and DS5000 running Ultrix 4.0
> (RISC)
> 
> (1) is there a public domain version of tex/latex around for these machines?

Sure; TeX has been running on the machines since before they were announced....
The UWash distribution should work fine.
> 
> (2) is there a public domain FORTRAN compiler available for these machines?
>     I looked at the subsets and did not find a fortran compiler for RISC
>     machines listed there even though there was a VAX fortran compiler
>     mentioned.

No public domain one I know of; we sell a Fortran compiler as a layered product.

> 
> (3) I intend to run NFS and YP on the two machines and since the /usr
>     file system occupies about 150 MB, and the 2100 and 5000 are binary
>     compatible, I wondered if there was a ``best'' way of saving space
>     without causing too many problems later. As far as saving space
>     is concerned, is it okay to share /usr between these 2 machines? If
>     this is not advisable, should I mount /usr/staff, /usr/users and
>     other directories individually? Is there any way of avoiding
>     duplication of files in the /usr file system?

The machines are fully binary compatible; different systems were only
required for basic machine support before Ultrix 4.0, which supports
all machines.  We run 5000's, 3100's, 2100's all from the same /usr areas here.
Unless you use YP elsewhere, I'd recommend using Bind/Hesiod for name services,
though.  YP is far from a wonderful name server...

    Thanks to Jim Gettys for this information.
                                            shirish.

grunwald@foobar.colorado.edu (Dirk Grunwald) (10/20/90)

I should have responded earlier.

I've been using the 'F2C' package available from research.att.com, and
then running the results through the C compiler on my DS3100. It's
cheaper the DEC fortran, and seemed to barf less often.  I've had
several programs in which 1.31 fortran died, but f2c -> cc1.31 didn't.
In the cases where f2c -> cc1.31 *did* die, running it through 'gcc'
gave a correct run, albeit 15% slower.

For programs that f77 *did* accept, I didn't see any performance
difference between f77 and f2c. MIPS made a good C compiler.

As for latex, it's out-of-the-box.