"Ben Voigt [C++ MVP]" <rbv@nospam.nospam> wrote in message
news:e0B5m9K8HHA.980@TK2MSFTNGP06.phx.gbl...
"Alexander Grigoriev" <alegr@earthlink.net> wrote in message
news:%237S4e$C8HHA.1212@TK2MSFTNGP05.phx.gbl...
Since it contains the 'using' directive, the original example was
compiled as C++. Anyway, in VC atof() is also declared in math.h
Absolutely.
I was trying to dispel Duane's confusion about what Barry wrote about
behavior with missing prototypes by clarifying that (a) it does not apply
to C++ and (b) what happens in C, which is admittedly tangential to the
original post where the proper prototype was in scope.
I was only confused WRT C and then only because
I didn't know we were talking about C. <g>
As it turns out, the original code was getting the prototype
for atof() from both math.h and via iostream so the
suggestion that started this was not correct. Which
pretty much means that we've extended this thread a bunch
without anyone answering the OP's question
(see the subject above<g>)
atof implementation. To benchmark vs2005's atof it would be far better to
compare to e.g. newlib's version using the gcc compiler, both running on the
same Windows OS and on the same CPU.