From 9a9faddbcb9f61a6dd527aacd41e18dc877649c7 Mon Sep 17 00:00:00 2001 From: Guido van Rossum Date: Tue, 10 Dec 1991 13:56:55 +0000 Subject: Formulate better error strings. --- Objects/floatobject.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/Objects/floatobject.c b/Objects/floatobject.c index 0966752..77015b9 100644 --- a/Objects/floatobject.c +++ b/Objects/floatobject.c @@ -174,7 +174,7 @@ float_div(v, w) return NULL; } if (((floatobject *)w) -> ob_fval == 0) { - err_setstr(ZeroDivisionError, "float division by zero"); + err_setstr(ZeroDivisionError, "float division"); return NULL; } return newfloatobject(v->ob_fval / ((floatobject *)w) -> ob_fval); @@ -192,7 +192,7 @@ float_rem(v, w) } wx = ((floatobject *)w) -> ob_fval; if (wx == 0.0) { - err_setstr(ZeroDivisionError, "float division by zero"); + err_setstr(ZeroDivisionError, "float remainder"); return NULL; } return newfloatobject(fmod(v->ob_fval, wx)); @@ -212,7 +212,7 @@ float_divmod(v, w) } wx = ((floatobject *)w) -> ob_fval; if (wx == 0.0) { - err_setstr(ZeroDivisionError, "float division by zero"); + err_setstr(ZeroDivisionError, "float divmod()"); return NULL; } vx = v->ob_fval; @@ -251,13 +251,13 @@ float_pow(v, w) return newfloatobject(1.0); /* x**0 is 1, even 0**0 */ if (iv == 0.0) { if (iw < 0.0) { - err_setstr(RuntimeError, "0.0 to the negative power"); + err_setstr(ValueError, "0.0 to the negative power"); return NULL; } return newfloatobject(0.0); } if (iv < 0.0) { - err_setstr(RuntimeError, "negative float to float power"); + err_setstr(ValueError, "negative float to float power"); return NULL; } errno = 0; -- cgit v0.12