[Info-vax] VSI OpenVMS Hobbyist Program Announced.

[Phillip Helbig undress to reply]

In article <gv7kehFtc50U1 at mid.individual.net>, Hans Bachner
<hans at bachner.priv.at> writes: 

> why not just provide new PAKs? I guess folks who have used *and 
> customized* the disk image would like to preserve their changes.
> 
> Of course they can by binding the new image as a separate disk to their 
> virtual Alpha, extract and re-apply the PAKs - but why do you want them 
> to take this extra effort?

Right.

In the longer term, there should be a supported way to do an upgrade 
without overwriting ANY customization.  Perhaps this would be as simple 
as (re)defining SYS$SYSROOT to have an additional translation before the 
other two, which could be a non-system disk, possibly shared by more 
than one node (or all nodes).  I've heard of people doing this, but I 
don't know to what extent it is supported.  Some things are relatively 
easy: have a line in SYLOGICALS.COM to mount this disk, then another one 
to execute a procedure there.  So moving SYSUAF etc off the system disk 
is relatively easy.  A bit more work for OPERATOR.LOG and the audit 
server, but still not that much work.  TCPIP is an additional burden, 
then things like procedures which, in contrast to SYSUAF, say, have no 
built-in logical names.  So I have stuff like a line at the top to look 
on the non-system disk for a procedure of the same name and execute it 
then exit, but these get overwritten by a system upgrade.  I've thought 
about customizing things to the max then copying only executables from a 
newer version to a system disk, but this is probably neither sufficient 
nor supported.