GNAT/RTEMS 4.5.0 build error in bit_ada

Joel Sherrill joel.sherrill at OARcorp.com
Sat Aug 23 13:33:02 UTC 2003


Ralf Corsepius wrote:

>On Sat, 2003-08-23 at 06:14, Dave Richards wrote:
>  
>
>>I am attempting to build GNAT/RTEMS from source on a Red Hat Linux 8.0
>>system (i386).
>>    
>>
>
>Urggs, I missed the subject "GNAT/RTEMS-4.5.0".
>
>Though GNAT/RTEMS-4.6 still has a lot of warts (Too much has changed
>about GNAT and RTEMS since 4.5.0 had been released [1]), I would not
>recommend to using GNAT/RTEMS-4.5.0
>  
>
Definitely.  From the testing I have done, GNAT as shipped with the 
current gcc series
is at the same level of useability as the older GNAT's based upon gcc 
2.8.1. 

It's been a long time since I ran bit_ada but it weas only designed for 
the GNAT
based upon gcc 2.8.1 and hasn't really been used since gnat 3.12p or 3.13p.
Building RPMs is just a lot easier and safer.

If there is no other reason to upgrade, using the newer GNAT shipped 
with GCC 3.x
supports a LOT more CPUs and is much better intergrated with the other 
languages.
For example, m68k-rtems simply did not work with the older gcc 2.8.1 
based GNAT
because that GCC did not support ELF on many embedded targets.

>Ralf
>
>[1] GNAT has been merged into gcc; RTEMS/GNAT has been merged into RTEMS
>as a GNAT-wrapper to RTEMS/c.
>  
>
GNAT/RTEMS was always a combination that used the C RTEMS as the 
run-time for GNAT.
Even when we got a compiler validation certificate, it was on this 
design.  Here's a link to it:

http://sw-eng.falls-church.va.us/ajpo_databases/ada95_validated_compilers.html#67

FWIW the behavior seems to be at least as good now.

--joel







More information about the users mailing list