[Info-vax] The future of Ada on VMS

[johnwallace4 at yahoo.co.uk]

On Monday, 21 May 2018 21:06:50 UTC+1, Simon Clubley  wrote:
> On 2018-05-21, Arne Vajhøj <arne at vajhoej.dk> wrote:
> > On 5/21/2018 1:38 PM, Dave Froble wrote:
> >> But once again, you're predicting without any idea of what might happen. 
> >>   You say there will not be an Ada compiler.  How do you know that? 
> >> Perhaps there is not one now, or any announced plans for one.  But VSI 
> >> will look at what their customers are asking for, and if there is demand 
> >> and good business sense to do so, do you doubt John's ability to produce 
> >> a new up to date Ada compiler?
> >
> > Ada is a bit more uncertain.
> >
> 
> Try a lot more uncertain (unfortunately).
> 
> > My understanding is that VSI has not talked about migrating
> > the extremely old DEC Ada 83 but has talked about getting ACT
> > to port their modern Ada. But ACT has dropped VMS as a platform,
> >
> 
> This is my understanding as well.
> 
> When DEC/CPQ/HP management decided to go with an external vendor
> for Ada instead of building an Ada 95 compiler in-house, that was
> when they lost control of the long term Ada ecosystem and became
> dependent on that external vendor's plans.
> 
> > I do not doubt that John could create an Ada compiler, but
> > every indication is that he will be working on other stuff.
> >
> 
> Creating an Ada compiler from scratch is a massive job. A number
> of people (who know way more about what is involved than I do)
> have looked at this possibility in various forms and have come
> to the same conclusion.
> 
> > So it will be VSI management talking to ACT management.
> >
> > Or talk to the great guys that made GNAT work on VMS I64
> > about an x86-64 version. See post some time ago!
> >
> 
> That's not really a meaningful comparison unfortunately.
> 
> The code for Itanium already existed in the code base he used and
> it was mostly a matter of learning how to use an existing GNAT Pro
> compiler for Alpha VMS to build an Itanium compiler.
> 
> All the necessary bits for x86-64 on VMS would need to be added
> to the code base before such an attempt could be made for x86-64.
> 
> This would mean however that VSI would be maintaining GCC for
> Ada only and maintaining LLVM for all the other compilers.
> 
> > If I were dependent on Ada on VMS, then I would also be
> > a bit concerned.
> >
> 
> Very much agreed.
> 
> Simon.
> 
> -- 
> Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
> Microsoft: Bringing you 1980s technology to a 21st century world

When you refer to "Ada on VMS", what (exactly) do you mean?

There are at least two possibilities (actually more, when the
various Ada specifications are considered, and other stuff too):

1) Ada compilation system (it's not just a compiler) running on VMS?
2) Ada runtime environment running on VMS?

Quite a few folks/organisations that use Ada don't necessarily
have the same host environment as runtime environment. ACT are
reasonably familiar with this arrangement; it's the traditional
host/target environment which embedded systems folks have used
for years, with varying degrees of success.

It's also similar to the approach used by VAXELN Ada (VAX as 
host and target, but VMS on host vs VAXELN on target).

And it's also similar to the approach used by XD Ada, or
whatever it's currently called, which was a derivative
of DEC Ada with VMS (on VAX, Alpha, or IA64) as host, 
and with M68K family as the runtime instruction set.

Those words of mine are probably not clear, and certainly 
not complete, so here's a chunk of Wikipedia's words on 
cross compilers:

"A cross compiler is a compiler capable of creating executable code for a platform other than the one on which the compiler is running. For example, a compiler that runs on a Windows 7 PC but generates code that runs on Android smartphone is a cross compiler.

A cross compiler is necessary to compile code for multiple platforms from one development host. Direct compilation on the target platform might be infeasible, for example on a microcontroller of an embedded system, because those systems contain no operating system." 
from
https://en.wikipedia.org/wiki/Cross_compiler

gcc is quite complicated (not necessarily in a good way), 
and anyone who has to build their own gcc toolchain has 
interesting times ahead of them, especially if they want 
the compilation system to run on something other than 
Linux. 

Which leads to the "Canadian cross compiler" concept. E.g. 
a compiler toolchain built on Linux-x86 host, intended 
for use on a Windows/AMD64 host, and intended to generate 
code to run on (say) an ARM/Android target system. The 
Wikipedia article mentioned above has more words on the 
subject, for anyone who's interested.

An Ada-compiler discussion talking about (e.g.) safety 
critical systems and compilation environments which doesn't 
clarify which options apply in the bigger picture is 
inevitably going to have lots of assumptions and mbiguities.

A possibly more important (but potentially less visible) 
discussion is how much money might be available to make 
some or all of these things happen.