[Info-vax] The future of Ada on VMS

[Simon Clubley]

On 2018-05-24, gérard Calliet <gerard.calliet at pia-sofer.fr> wrote:
> It is strange for me also, but I could not event have a meeting at the 
> boot camp (where I was sponsor in the name of VMS ADA ALLIANCE).

Before replying to the following in detail, let me make something clear.

It is well known around here that VSI (especially VSI management) and
I do not agree on a number of things. However, I strongly agree with
everything that VSI writes below and I think the VSI position is very
reasonable and understandable when you look at the bigger and long term
picture.

> The prerequisite VSI told me are here:
> >>>
> What is missing are the following.
>
>      A complete development environment that includes not only the 
> compiler but OpenVMS style debugging, predefined libraries for OpenVMS 
> style language support (sorting, math, string manipulation, POSIX, XML 
> processing, Web integration, multi-language support, etc.).
>

The predefined libraries stuff is very important. This is so important
that back in the days of a freely available GNAT compiler for VMS
(about 15 years ago), Compaq used to offer their own predefined libraries
package that you could freely download (after registering) to install on
top of the Adacore compiler.

The list above just reflects how things have moved on in the last 15 years.

>      Installation testing tools like an IVP (Installation Verification 
> Procedure) that is customer viewable.
>

Absolutely standard for any VMS product.

>
>      Certification/validation tools that are VSI viewable only for 
> certifying the compiler adheres to Ada standards.
>

Absolutely expected for anything VSI are going to market on your behalf,
especially something as mission critical as an Ada compiler, given
where it is likely to be used.

They need to know that your product meets certain standards and that
you can prove this. If VSI market a lousy product, this will reflect
badly on them.

BTW, did ACATS show any areas of concern in your newly built Itanium
compiler ?

>      Installation kit.  An Ada ?kit? will have to be developed. Today 
> the Ada compiler is distributed as executable files the customer has to 
> install manually.  An installation procedure needs to be developed to 
> install the product in the correct directories for ease of use and for 
> product verification purposes.
>

Absolutely standard for any VMS product.

>      Documentation like the Installation Guide, Run-Time Library Users 
> Guide, Language Reference Manual, OpenVMS Extensions, Debugger Reference 
> Manual, Release Notes and other related documents.  These manuals will 
> have to be updated over time to reflect newer language standards and 
> enhancements to OpenVMS that may affect the Ada environment.
>

Absolutely standard for any VMS product. Perhaps you can start with the
existing GNAT documentation if the licence terms allow this.

>
>      Support, beyond just bug fixes, is a big issue as the compiler will 
> have to be kept up to date with the latest standards, new test routines 
> will have to be developed to test these new language standards and 
> product bugs will have to be fixed.  This will require a development, 
> test and support staff for this to be viable.
>

Absolutely standard for any VMS product. You need to demonstrate to VSI
that you can fix any problems which are thrown your way and that you can
fix them on certain timescales. It's an Ada compiler which means that it
is going to be used for Really Important Stuff. There are certain standards
expected when you supply tools for Really Important Stuff.

You need to demonstrate that you have a minimum level of infrastructure
to support your compiler and that in a year's time you are not going to
get bored and move onto something else.

Even serious volunteer hobbyist projects have this level of infrastructure.
The following is an experimental OS used for research and teaching that
I have been looking at and plan to play with:

http://www.helenos.org/

Look at all the infrastructure they have for what is just a university
and general research project. Now imagine the level of support that VSI
would expect you to be able to provide for something as important as
a commercial Ada compiler.

>      IP license rights.  Who owns the IP rights to not only the compiler 
> but other tools and OpenVMS modifications made to the Ada compiler? 
> These issues have not been worked out between you and David Sauvage. 
> This in itself is a big risk for VSI.
>

This will be a stonking massive, massive issue for VSI and rightly so.

It is also one of the first questions I asked you when you announced
this port because I wanted to be sure that you hadn't done something
that would lead to GPL contamination (for example) for anyone using
your compiler.

You need to be able to show the exact history of your port, starting
with the licence terms of your starting compiler, showing exactly
where you got any code you added to your port and its licence, and
that you have the rights you claim you do to any finished product.

This is also an absolutely standard business requirement.

> >>> end of quote
>
> It is not something I can do with my means. If you know an investor, 
> perhaps, I could reopen the discution with VSI :)

Simon.

-- 
Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world