[Info-vax] New CEO of VMS Software

[Dan Cross]

In article <unck70$p3mp$4 at dont-email.me>,
Lawrence D'Oliveiro  <ldo at nz.invalid> wrote:
>On Sat, 6 Jan 2024 15:59:58 -0500, Arne Vajhøj wrote:
>
>> No such concern was made for the Ultrix customers going to DEC OSF/1 aka
>> DUNIX aka Tru64.
>> 
>> DEC made less money from Ultrix. Ultrix and OSF/1 was two different
>> Unixes so compatibility would have been difficult anyway. And porting C
>> code using a C API was easier anyway.
>
>You almost got the point, didn't you? That POSIX had defined standard 
>types like "time_t" and "size_t", and code that was written to adhere to 
>those types as appropriate was much easier to port between different 
>architectures. This applied to customer code, to third-party code ... to 
>all code.

It took literally decades from the introduction of 64-bit Unix
machines until most software was 64-bit clean.  I was there; it
was a painful time, and Linux was actually behind the curve
here compared to many of the commercial vendors.

The mere existence of those types a) didn't help the piles of
code that was sloppy and made assumptions about primitive types
and b) didn't help with binary compatibility during the
(lengthy) transition period.  And yes, binary compatibility
mattered to a lot of people.

>And POSIX already existed when Dave Cutler commenced development on 
>Windows NT. Back when he was starting VMS, he could claim ignorance of 
>such techniques for avoiding obsolescence; what was his excuse this time?

What was Linux's?

>> VMS has only 64 bit code but both 32 bit pointers and 64 bit pointers
>> (32 bit pointers getting extended to 64 bit addresses).
>
>Not sure how you can have 64-bit code without 64-bit addressing ...

Of course you're not.  "64-bit code" for something like x86
refers to details of the processor mode and e.g. the handling
of the REX prefix.  On Alpha or Itanium, presumably that means
using the 64-bit ISA that uses e.g. 64-bit registers and so on.

But in either case, that's distinct from data pointers in
userspace are truncated represented as 32-bit values, as only
the low 2GiB of the address space is used by VMS applications.

>that is practically the essence of 64-bit code.

Nope.

	- Dan C.