obnoxious "not implemented" messages

[Ulrich Drepper]

Richard Henderson <rth@cygnus.com> writes:

> Why don't we return SNaN instead?  I personally hate libraries that
> print stuff.  Especially when there's no way to turn it off.

You will not see the message unless your program is really broken.

The only reason Zack cares about this is since his test suite will see
the messages when looking at the results from the libm-test.c use.  No
working program ever will use the functions.  And if a program really
uses them I want to know why it breaks and not have a rather cryptic
diagnosis via a SNaN.

-- 
---------------.      drepper at gnu.org  ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Cygnus Solutions `--' drepper at cygnus.com   `------------------------