[Info-vax] link to what is actually being ported to x86?

gérard Calliet gerard.calliet at pia-sofer.fr
Thu Jun 6 01:51:56 EDT 2019


Le 05/06/2019 à 00:52, Arne Vajhøj a écrit :
> On 6/4/2019 6:31 AM, Ian Miller wrote:
>> On Monday, June 3, 2019 at 6:39:24 PM UTC+1, Simon Clubley wrote:
>>> On 2019-06-03, Ian Miller <gxys at uk2.net> wrote:
>>>> On Monday, June 3, 2019 at 1:26:36 PM UTC+1, Simon Clubley wrote:
>>>>> On 2019-06-02, Dave Froble <davef at tsoft-inc.com> wrote:
>>>>>> On 6/2/2019 7:16 PM, seasoned_geek wrote:
>>>>>>> We've had many a discussion, but is there an official link to 
>>>>>>> what is and is not being ported to x86? I know we had 
>>>>>>> conversations about "desktop" and got some "server only" answer 
>>>>>>> but that didn't really answer the what is/is not since there has 
>>>>>>> to be some kind of console/terminal even for a server.
>>>>>>>
>>>>>
>>>>> Ada will not be available on x86-64.
>>>>>
>>>> people are still working on the ADA compiler question. It's 
>>>> important to some companies.
>>>
>>> Interesting.
>>>
>>> I wonder if people are more interested in running old Ada 83 code,
>>> in which case I wonder if it's viable to port DEC Ada across, or
>>> if they need Ada 95 and above, in which case they need GNAT or
>>> another Ada 95 and above compiler.
>>
>> I expect any future compiler will be based on the GNAT ADA compiler 
>> starting with the opensource version. I am told that the ADACore port 
>> of this to OpenVMS I64 does compile old ADA83 sources.
> 
> If Ada95 is a superset of Ada83 then it has to.
> 
> Arne
The GNAT Ada compiler I built from sources of course can compile Ada 83.
I used it to compile projects that used DEC Ada on Alpha. The fun was 
taht DEC Ada, which is said Ada 83, had already some features of Ada 95. 
So, in my project, all the sources were compiled with /83 and when 
rejected, I escaladed in /95 . Works fine.

The Gnat Ada I built is based on gcc 4.7 .Newer versions were more 
difficult to built, because of the choice of gcc to build on C++, and 
the fact Adacore gave up its work on VMS at that time.

(The Gnat Ada gcc build is a canadian cross compiled method from Linux).

So the version of GNAT Ada I have is a little bit old. And I don't know 
if the newer versions maintain /83 compilation.

Anyway the future of Ada on VMS is not DEC Ada. I concluded that hearing 
what John Reagan told about it.

For Itanium, it seems we will be able to have more recent builds from 
gcc sources, when VMS itanium will get C rtl improvements.

For x86 we will be able to do the same cross compile buil from gcc. And 
because Adacore works on a LLVM front end prototype, we'll be able to 
use it to have a native Ada compiler on x86 / LLVM.

For Alpha, we 'll have the choice to improve or just maintain the DEC 
Ada, or try a cross compile build whith gcc, or wait to the 'LLVM 
building GCC front something' method I heard about. But perhaps the site 
which are still on alpha don't need any modernization of their Ada 
applications.

One important aspect here is the information. As for VMS it is important 
to say "there is a clear future for Ada on VMS, and there is now support 
for it". Because this information inhibit somehow the wrong decisions to 
go out from VMS.




More information about the Info-vax mailing list