[Info-vax] The Open-Sourcing Discussion, Yet Again

[johnwallace4 at yahoo.co.uk]

On Wednesday, 11 December 2013 21:48:29 UTC, Bob Koehler  wrote:
> In article <52a8d982$0$1408$c3e8da3$b1356c67 at news.astraweb.com>, JF Mezei <jfmezei.spamnot at vaxination.ca> writes:
> 
> > 
> 
> > No, the whole point of making VMS open source is to allow bits to be
> 
> > ported to other operating systems. Consider for instance moving TPU to
> 
> > OS-X/Linux.
> 
> 
> 
>    TPU was already cloned to other platforms by a/Soft.  Check and see
> 
>    how profitable it was for them.
> 
> 
> 
>    And they didn't port DEC's BLISS code, they wrote it from scratch.
> 
>    You want to do a port, you get to help GNU finish thier BLISS
> 
>    compiler.

"You want to do a port, you get to help GNU finish thier BLISS compiler."

Sounds good to me, but then what do I know (apart from the fact that compiler 
development requires a mindset which I don't have, and can offer a salary which 
I'd love to have).

Same approach may sort Ada too, if Ada really is needed, if GNAT's good enough 
for safety critical stuff that flies (and the regulators so far seem to think 
it is).

I can't remember where VMS on IA64 got to with object file formats, 
linker, libraries, debugger, etc. Details (but far from trivial. And 
important details, both to "VMS engineering" and, in time, to end customers).

Many years ago, VMS may have needed a complex custom build environment. 
Certainly RCS and SCCS and the original make weren't up to much when I was on a 
50+ person three-country three-OS team in the 1980s.

Nowadays, there's git and other tools which appear to work with decentralised 
development. A different IDE for every day of the month. Make tools which 
on the whole seem to work, and stuff like Jarvis to glue it all together (and python or even DCL for simpler stuff).

Some of the relevant open source bits are already available on VMS, and the 
good people in the VMS open source programme are hopefully working on others. 

Millions of users voluntarily using these tools can't all be wrong, can they?

Note that in this imagined world I'm not *assuming* open source. Open source is 
obviously one option (and potentially a commercially attractive one to end 
customers).

A commercial transfer, iirc as was done with VAXELN when DEC's Embedded and 
Realtime folks were sold off, is another option. Unfortunately for VAXELN 
customers the product was already on its last legs (the hint is in the name). 

HQ had previously made the decision that hard-realtime VAX customers would 
migrate off VAXELN to a DEC-badged, but largely incompatible with VAXELN, Alpha-
targeted version of Wind River's VxWorks. Damned awkward customers didn't 
actually want to throw away their investment in software and systems and 
experience, can you imagine that! Can you imagine such resistance happening 
*multiple times* to the same customers of the same company with multiple 
generations of hardware and software, and the message still doesn't sink in!

Have a lot of fun.