Re: Drop VMS?

["Zack Weinberg" <zackw at Stanford dot EDU>]

On Tue, Feb 27, 2001 at 05:51:41AM -0500, dewar@gnat.com wrote:
> <<- Previous inquiries about the existence of a GCC user community on
>   VMS have turned up completely blank.
> >>
> 
> GNAT is ported to Alpha VMS, and we have a significant user community.
> Furthermore, we anticipate no specific problems in going to GCC 3.0
> for this port, and are planning to do that in the very near future.
> We owe some patches in this area, and will try to get these checked
> in in the next few weeks.

This is good to know.  I still have serious concerns about the health
of the language-independent code and C front end.  For instance, no
one has bothered to update makefile.vms in ages; the configuration
parameters are hardwired in vms.h and xm-vms.h, and appear to be wrong
(this is the stuff that autoconf detects on Unix hosts); the pathname
parsing code doesn't have a clue how to do VMS filename syntax; etc.
But if it's under active maintenance in your tree, the problems
shouldn't be that hard to resolve.

Part of the problem is I wouldn't have the faintest idea how to
configure and build GCC on a VMS machine, and not just because I don't
speak DCL.  All the installation documentation assumes the configure
script works; when there's no Bourne shell to run it with, well...
(Someone must have *thought* it worked, because there's mention of
{alpha,vax}-dec-vms in config.gcc.  Color me confused.)

> I certainly agree on dropping DEC VMS, but Alpha VMS is alive and well
> as far as we are concerned.

Did you mean VAX here?  The Alpha VMS port is a good deal less crufty
than the VAX port, indeed.

zw