karl@haddock (09/30/86)
orion!vaughan writes: >After having my code bomb several times (unfortunately, I was not as lucky >as you - instead of zeroes, I got core dumps), I discovered this ... That's unlucky? I'd rather get a core dump (which tends to point right at the error) than the wrong answer (which may not become visible for a while). Karl W. Z. Heuer (ima!haddock!karl or karl@haddock.isc.com), The Walking Lint
karl@haddock (09/30/86)
ucsfcgl!kneller (Don Kneller) writes: >In article <22@orion.UUCP> vaughan@orion.UUCP (Robert Vaughan) writes: >>scanf("%lf",&foo); >Except, of course, in the infinite wisdom of Silicon Graphics whose >floats and doubles are the same size so one must use %f for both. The >"justification" is that they have yet another floating point type called >long float with gets the %lf format. It's painfully nonportable. Sigh. Then it isn't a C compiler and/or library. Do you really mean "long float"? In K&R that's a synonym for "double"; ANSI removed this and added "long double" for which the format is "%Lf" (note uppercase L). The elegant solution would've been "short float" (formerly "float"), format "%hf"; "float" (formerly double), format "%f"; and "long float" (what ANSI calls "long double"), format "%lf", but elegance must defer to compatibility. Sigh. Karl W. Z. Heuer (ima!haddock!karl or karl@haddock.isc.com), The Walking Lint