[Info-vax] The future of Ada on VMS
Simon Clubley
clubley at remove_me.eisner.decus.org-Earth.UFP
Thu May 24 16:23:27 EDT 2018
On 2018-05-24, gérard Calliet <gerard.calliet at pia-sofer.fr> wrote:
> Wrong story, Simon. I never got the GNAT Ada pro, because Adacore didn't
> want to give it anymore. VSI had a pcsi GNAT Ada pro, and at that time
> they didn't yet unpack it, and didn't do anything for us and our
> customer (first VSI license in the world).
> [[About all the discution about VSI: you are totaly right, they are
> totaly right. And Digital, Compaq, HP, HPE have always been totaly
> right. I know it.]]
>
Interesting. I thought you had mentioned doing it by starting with
GNAT Pro. My mistake, sorry.
> So we (AdaLabs, me) did rebuild the compiler from FSF sources,
> integrating some bits of code specific to VMS from "friends" (very
> little) and patching some code (very little). It is a canadian
> cross-compilation operation.
>
What are the licence terms for those bits of code ?
> The limitation is that we used an old gcc because it was much simpler.
Now that's where we took a very different approach, I wanted the
latest Ada compiler so I had to try and deal with the C++ problem
from gcc 4.8.0 onwards. I did that by taking the Linux based cross
compiler approach.
I was also targeting VMS Alpha and it looked to me like the code
for VMS Itanium was in better shape than that for VMS Alpha in the
gcc and binutils versions I looked at.
> But when the built is done, IT IS the gnat Ada Pro : Adacore has to be
> compliant to the license Open Source, so the sources in FSF ARE the
> sources of Gnat Ada pro. And, of course, a lot of context tools can be
> ported, but we didn't get time and/or investment.
>
No they are not the same sources. Adacore maintains its own branch of
the GNAT source code which it uses for GNAT Pro. Every so often, it
pushes changes from this tree into the FSF tree but it is not actually
legally required to push anything.
If you read the GPL, you will see that Adacore are only required to
supply its customers (and only its customers) with the source code
for the compiler those customers have bought. They are not required
to push everything into the FSF code base.
The code bases are not identical. For example, the number of targets
supported by GNAT Pro is larger than the number of targets supported
by FSF GNAT.
Did ACATS pick up any issues or missing features in your Itanium compiler ?
Also, I have a good reason to ask about GPL contamination. As well as
GNAT Pro and FSF GNAT, there is also the Community Edition directly
available from the Adacore website.
However, any user programs compiled using this version of GNAT fall
under the GPL because the RTL in this version of GNAT is pure GPL.
This means that if you took anything from the RTL in the Community
Edition of the GNAT compiler available from the Adacore website then
that code is pure GPL and so is anything that it is inserted into.
Unlike FSF GNAT, there is no runtime exception for the Community Edition
of the Adacore GNAT compiler.
Yes, I know it's a real mess that only makes things confusing. :-( :-(
And yes, there has been much discussion about this over in the Ada world.
> With a little effort, everything is in FSF sources. I work now on dwarf
> issues. Just beginning. I'll say more later.
>
Simon.
--
Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world
More information about the Info-vax
mailing list