[Info-vax] Fortran

[Phillip Helbig undress to reply]

In article <pu561b$k0q$1 at dont-email.me>, Dave Froble
<davef at tsoft-inc.com> writes: 

> >> Your claim about how hard it will be to move is just that, your claim.
> >> I feel that you are mistaken.  But you are entitled to your opinion.
> >
> > That obviously depends on the system.  Some will be easier, some less
> > so.  Some will benefit greatly from a mixed-architecture cluster, some
> > less so.
> >
> >> I'm rather sure that x86 VMS will still mount disks, should the
> >> interface be the same, from VAX, Alpha, and that sinking ship.
> >
> > So x86 machines running VMS will have SCSI interfaces?
> 
> Don't know, we'll see.

If they don't, then mounting the disks I have is not an option.

> >> I'm even more sure that BACKUP will be able to take images from those
> >> disks and restore them to disks on an x86 running x86 VMS.
> >
> > Sure.  But that won't work out of the box for all users.
> 
> When hasn't VMS BACKUP not worked?

Backup itself, sure.  But in at least one file, some procedure will have 
hard-coded disk names, if only to define logical names.  Will these be 
the same?  I don't know.

> >> I'm guessing that x86 VMS will be able to have network links to Alpha
> >> and that sinking ship, and files, and save sets, could be copied, and /
> >> or restored.
> >
> > Sure.  But that won't work out of the box for all users.
> 
> When hasn't networks not worked on VMS?

Well, for one thing there have been many bugs in various TCPIP 
implementations.  But, again, that's not my point.  If it were that 
easy, the VAX-to-ALPHA transition would have been "shut down, mount the 
disk on ALPHA, reboot".  While each hardware transition probably become 
easier, it is never an out-of-the-box situation.

> Everything will need to be compiled and linked on x86.  Unless you're 
> using one of the VEST type of things.  I fail to see how mixed anything 
> has a thing to do with the compilers and linkers.

There is a system which works.  I have to compile, link AND TEST new 
code on the new system.  Do I want to completely shut down the old 
system and troubleshoot on the new one?  What about comparing 
executions.  And so on.

There is a reason why mixed clusters for migration exist.