Re: Is there really any interest in a patch to allow cvs 1.11.6

[Paul Edwards]

"Derek Robert Price" <derek@ximbiot.com> wrote in message 
news:mailman.1329.1065623934.21628.bug-cvs@gnu.org...
> Is there a good reason, outside of the C89 spec only guaranteeing
> EXIT_SUCCESS & EXIT_FAILURE, for:

Actually, 0 or EXIT_SUCCESS.

> but does anyone know any hosts which, in practice, return some value
> other than `unsuccessful termination' or the actual STATUS when STATUS
> is a number in the range 2 through 255?

1 is not well-defined under MVS.  Normally, 0 is success, 4 is
warning, 8 is error, 12 is severe error, 16 is catastrophic error.

I think EXIT_FAILURE is generally set to 8.

I think VMS has some scheme where even numbers are
errors, odd numbers are OK, so although I don't recall
seeing it implemented that way, I would expect to see
0 get translated into 1 on the way out.

The C standard is written that way because it was the best
compromise available.

> A similar configure test could quite easily tell sanity.sh's
> dotest_status function that any non-zero return value it is told to
> expect will actually be $EXIT_FAILURE on offending systems.

I think the current C code is fine, ie it conforms to C89.
But if a non-default option called:

ALLOWS_COMMON_RETURN_CODES

was to be defined, that passed err unchanged to exit, that would
be fine by me too.

BFN.  Paul.