[Info-vax] What would you miss if DECnet got the chop? Was: "bad select 38" (OpenSSL on VMS)

[Stephen Hoffman]

On 2016-09-20 10:00:24 +0000, Chris said:

> Which is why, once again, VMS needs a software abstraction layer that 
> presents a unix  / Linux interface.

If you're suggesting work on some sort of software abstraction layers 
and something beyond C11 and maybe some related POSIX giblets and 
related support, then I'd rather see the OS API provide an OO 
interface, and let the higher layers — higher layers either from VSI or 
from open-source or end-users — deal with what Unix and Linux 
applications might expect.  If there's going to be effort expended in 
this area to abstract the operating system interfaces, that is.   
Because if I wanted ("just") what Unix and Linux provided at the API — 
past C11 et al — then I'd probably be better served by just using Unix 
or Linux.   If you're going to invest, then providing a fundamentally 
more capable and/or better and/or better abstraction than what (most) 
Unix systems offer provides end-users and ISVs with more capabilities 
and more flexibility.

> Do that once, do it right and most if not all of the useful os code is 
> available with just a recompile.

Most code.  There are APIs that OpenVMS likely won't ever do well very 
with, however.   The fork() being key among those and there are issues 
with SSIO, but there are others that would be problematic, and there 
are more than a few APIs that are just missing.

> A short step from that to a full package oriented environment for VMS.  
>  Sheer luxury :-)...

So long as we're (also) no longer chasing around itemlists and patch 
downloads and the rest of the baggage, and if we can finally toss those 
blasted decc$ logical names and related into the dustbin of history, 
sure.


-- 
Pure Personal Opinion | HoffmanLabs LLC