[HECnet] long term paks

[Thomas DeBellis]

I think that's an easy case to make for hardware; sooner or later you 
run out of vacuum tubes...

Less so with software or maybe not at all.  Today, at work at Big Bank, 
I can sign on to TSO and it hasn't changed since I was a freshman, eons 
ago.  That's annoying because I never really did care for 3270's.  
However, on the exact same machine, once you are on, you can switch into 
Linux (I.E., the OMVS segment).  So you also get all the rotten cryptic 
utilities that annoy me...

But the children like it because that's what they know.  So you can keep 
developing in C, Java, C++, Python or whatever is sexy this decade.  You 
also don't lose your bajillions of hours of man years in COBOL (which 
really is easy to learn).  It's not just whether there is anybody around 
who understands the software, it's also the cost to port all of it.  
Even extending year fields by 16 bits turned out to be prodigiously 
expensive.

But the value add of the VMS x86_84 is that the base _architecture_ is 
very unlikely to ever go away.  Not that new hardware won't happen, but 
that you can't realistically run out of address space.  The main driver 
that kills architectures is address space; not what the processor can 
compute, but address space.  z hardware is 64 bit; IBM kept coming up 
with ways to extend the virtual address space. 24-> 31-> 64.

You can also choose not to make it anymore because you can't compete; 
this is what happened with Alpha, the manufacturing costs essentially 
because overwhelming as they tried to keep up with Intel/AMD (think die 
size)

On 3/14/20 12:35 AM, David Moylan wrote:
> All technology has a use by date at which time it’s only of interest 
> to enthusiasts, hobbyists, collectors and the like. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sonic.net/pipermail/hecnet-list/attachments/20200314/4e1e9e5f/attachment.html>