Pedro Alves wrote:
> On Friday 23 October 2009 17:00:20, Michael Snyder wrote:
>> Pierre Muller wrote:
>>> Is there a valid reason to use LITTLE_ENDIAN rather than
>>> BFD_LITTLE_ENDIAN as required by the ARI rule:
>>>
>>> LITTLE ENDIAN 3       Do not use LITTLE_ENDIAN, instead use
>>> BFD_ENDIAN_LITTLE
>>> See:
>>> http://sourceware.org/gdb/current/ari/
>>>
>>> Pierre Muller
>>> as ARI "maintainer"
>>
>> Thanks Pierre.  Checked in as obvious.
> 
> No, this is wrong.  BYTE_ORDER isn't available on all hosts, and
> even if it did, its values surely don't match enum bfd_endian...

Dang.  How about this?

The byte order that's saved in the file is never actually used.
We just need to make sure that it's a consistent order, and that
hopefully we can read the execution log into gdb on a different
host.  So target byte order should be all that matters.