On Fri, Feb 25, 2011 at 1:02 AM, Mark Overmeer <mark@overmeer.net> wrote: > * Steve Peters (steve@fisharerojo.org) [110225 04:05]: > > On Tue, Feb 22, 2011 at 10:27 AM, Robert Lang <perlbug-followup@perl.org > >wrote: > > > $ perl -MPOSIX -e 'print isdigit(""), "\n"' > > > Thanks for the report. I do have a fix ready for this, but before I > commit > > this, this technically effects all the ifXXX() functions. Also, in the > > ext/POSIX/t/posix.t we have... > > The change in C is simple, of course, but it will break existing > implementations. This is a flaw in the docs, not in the code. > Occasionally, people depend on the behavior of bugs, like people that depending on Carp being accidentally pulled in with "use warnings". That makes it no less of a bug to be fixed. As the behavior of the function is not documented regarding what happens when an empty string or undef is passed in, we still have room to define it. I'll be sure to add the documentation fix as well. Also, looking into the history of the code, it did at one time do exactly what Linux did when passed an empty string. It core dumped. Steve Peters steve@fisharerojo.orgThread Previous | Thread Next