This is the mail archive of the
libc-hacker@sourceware.cygnus.com
mailing list for the glibc project.
Re: obnoxious "not implemented" messages
- To: drepper@cygnus.com (Ulrich Drepper)
- Subject: Re: obnoxious "not implemented" messages
- From: Zack Weinberg <zack@rabi.phys.columbia.edu>
- Date: Tue, 28 Jul 1998 13:25:59 -0400
- cc: libc-hacker@cygnus.com
On 28 Jul 1998 10:17:33 -0700, Ulrich Drepper wrote:
>Zack Weinberg <zack@rabi.phys.columbia.edu> writes:
>
>> The stub long-double math functions print an obnoxious message in
>> addition to returning ENOSYS. Why is this done? It doesn't help the
>> user any; it can cause crashes (e.g. if I close stderr); and it
>> collides with expectations that the math functions have no side
>> effects.
>
>In this cae it's better to have a crashing program than one which does
>not test for plausible results. If somebody uses the long double
>float functions on a platform which does not support them everything
>can happen and I think it is better to make this very clear, not only
>while compiling where messages can get lost in the bulk of compiler
>messages.
I bring this up because the messages confuse my new test suite, which
expects that test cases will produce very specific output. libm-test
does the right thing with the unimplemented functions, so there is no
problem in this case; shouldn't there be a way to shut the errors off?
zw