[uClibc] __isnanl and __isinfl #ifdefed out?
Rob Landley
rob at landley.net
Mon Nov 29 03:35:15 UTC 2004
On Sunday 28 November 2004 10:27 pm, Erik Andersen wrote:
> On Sun Nov 28, 2004 at 09:17:41PM -0500, Rob Landley wrote:
> > Why are there #if 0 guards around these functions in
> > uClibc/libm/fpmacros.c? math.h happily makes calls to the these suckers
> > for isnan() and such when used on a long double, and then it dies at link
> > time. It might be nice to either take 'em out of the headers, or put 'em
> > backin the library. (Maybe a
>
> Recall that TODO item about auditing header files? You've
> found a prime example of that task not having not yet been
> completed,
>
> -Erik
Glad to be of service. I'd happily fix it if I had the foggiest idea why the
#if 0 was in there in the first place. (Was it chopped out because it
doesn't work? Because it took up too much space? Should it be
#if CONFIG_DO_C99_MATH instead of #if 0? Shouldn't strtold() be chopped out
also if long double support is missing?)
I do intend to help audit the headers, I'm just working through my busybox
to-do list first. (The next low-hanging fruit was "sort", that lets me get
rid of coreutils entirely in my development environment and
still ./configure;make;make install everything except perl, which needs
"comm". I've already upgraded my little distro to a CVS snapshot of uclibc,
and this problem was from that rather than 0.9.26.)
Rob
More information about the uClibc
mailing list