Commit 8767d071 authored by Damien George's avatar Damien George
Browse files

py: complex_print uses format_float if single precision fp used.

parent bee17b00
......@@ -12,6 +12,10 @@
#if MICROPY_ENABLE_FLOAT
#if MICROPY_FLOAT_IMPL == MICROPY_FLOAT_IMPL_FLOAT
#include "formatfloat.h"
#endif
typedef struct _mp_obj_complex_t {
mp_obj_base_t base;
mp_float_t real;
......@@ -22,11 +26,24 @@ mp_obj_t mp_obj_new_complex(mp_float_t real, mp_float_t imag);
STATIC void complex_print(void (*print)(void *env, const char *fmt, ...), void *env, mp_obj_t o_in, mp_print_kind_t kind) {
mp_obj_complex_t *o = o_in;
#if MICROPY_FLOAT_IMPL == MICROPY_FLOAT_IMPL_FLOAT
char buf[32];
if (o->real == 0) {
format_float(o->imag, buf, sizeof(buf), 'g', 6, '\0');
print(env, "%s", buf);
} else {
format_float(o->real, buf, sizeof(buf), 'g', 6, '\0');
print(env, "(%s+", buf);
format_float(o->real, buf, sizeof(buf), 'g', 6, '\0');
print(env, "%sj)", buf);
}
#else
if (o->real == 0) {
print(env, "%.8gj", (double) o->imag);
} else {
print(env, "(%.8g+%.8gj)", (double) o->real, (double) o->imag);
}
#endif
}
STATIC mp_obj_t complex_make_new(mp_obj_t type_in, uint n_args, uint n_kw, const mp_obj_t *args) {
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment