[Info-vax] Making the CRTL version dependency information useful

[Bill Gunshannon]

On 01/16/2018 08:21 AM, Simon Clubley wrote:
> On 2018-01-15, johnwallace4 at yahoo.co.uk <johnwallace4 at yahoo.co.uk> wrote:
>> On Monday, 15 January 2018 16:19:48 UTC, Ian Miller  wrote:
>>>
>>> There was a OpenVMS ADA from Adacore but they stopped supporting it.
>>> There is this one http://www.vmsadaall.org/index.php/en/
>>>
>>> I hope VSI are talking to someone about ADA as there is definitely a need for a support OpenVMS ADA compiler
>>
>> Do you have a definitive reference for "they stopped supporing
>> it"? There are signs from 2017 that although 'it' may have gone
>> away from the AdaCore website, 'it' can still be done, with or
>> without involving AdaCore.
>>
> 
> Yes and No.
> 
> I was never able to get an Ada compiler built for VMS Alpha using
> the pure FSF sources. The compiler you refer to appears to be have
> been built using the GNAT Pro compiler as the starting point which
> is an option not available to me and would make the process
> considerably easier and more viable.

I started with GNAT from the first release from NYU.  The chain
of development required that you use the previous release to
build the next in the chain.  After ACT took it over the chain
was broken and some of the missing versions were not made available.
Thus, you need an Ada compiler to build the current Ada compiler.
So much for the "protection" of IP provided by the GPL.

> 
> The issue is that GNAT is written in Ada which means you need an
> Ada compiler to compile it.

Exactly, see comments above.  If you were keeping up with developments
when NYU was doing it that wasn't a problem.  But, today, there are
pieces (deliberately?) missing.

> 
> The approach I took was to try to build a cross compiler gcc/binutils
> toolchain on Linux for a VMS Alpha target. Using this approach, it
> looks like either various bits might be missing or there are some
> non-obvious steps needed which I have missed.
> 
> I should point out that I have built gcc cross compilers from source
> a good many times and for a range of targets so the general process
> is well known to me.
> 
> While I was not able to get Ada built, I was able to get a C cross
> compiler built using the same method and binaries produced using
> this Linux based cross compiler ran on VMS Alpha.

I would imagine that if you gathered up all the original NYU work
and started from the very beginning you wold reach a point after
ACT took over where the previous version would not build the next
version thus making it impossible for anyone but ACT to continue
the chain.

bill