Commit 071a71a8 authored by Piotr Caban's avatar Piotr Caban Committed by Alexandre Julliard

msvcrt: Remove unused parameter in pf_integer_conv.

parent 704d0662
......@@ -514,8 +514,7 @@ static inline int FUNC_NAME(pf_output_hex_fp)(FUNC_NAME(puts_clbk) pf_puts, void
/* pf_integer_conv: prints x to buf, including alternate formats and
additional precision digits, but not field characters or the sign */
static inline void FUNC_NAME(pf_integer_conv)(APICHAR *buf, int buf_len,
pf_flags *flags, LONGLONG x)
static inline void FUNC_NAME(pf_integer_conv)(APICHAR *buf, pf_flags *flags, LONGLONG x)
{
unsigned int base;
const char *digits;
......@@ -783,7 +782,7 @@ static inline int FUNC_NAME(pf_output_fp)(FUNC_NAME(puts_clbk) pf_puts, void *pu
limb_len = LIMB_DIGITS;
}
radix_pos -= f.Precision;
FUNC_NAME(pf_integer_conv)(buf, ARRAY_SIZE(buf), &f, l);
FUNC_NAME(pf_integer_conv)(buf, &f, l);
r = pf_puts(puts_ctx, f.Precision, buf);
if(r < 0) return r;
......@@ -824,7 +823,7 @@ static inline int FUNC_NAME(pf_output_fp)(FUNC_NAME(puts_clbk) pf_puts, void *pu
limb_len = LIMB_DIGITS;
}
prec -= f.Precision;
FUNC_NAME(pf_integer_conv)(buf, ARRAY_SIZE(buf), &f, l);
FUNC_NAME(pf_integer_conv)(buf, &f, l);
r = pf_puts(puts_ctx, f.Precision, buf);
if(r < 0) return r;
......@@ -870,7 +869,7 @@ static inline int FUNC_NAME(pf_output_fp)(FUNC_NAME(puts_clbk) pf_puts, void *pu
limb_len = LIMB_DIGITS;
}
prec -= f.Precision;
FUNC_NAME(pf_integer_conv)(buf, ARRAY_SIZE(buf), &f, l);
FUNC_NAME(pf_integer_conv)(buf, &f, l);
r = pf_puts(puts_ctx, f.Precision, buf);
if(r < 0) return r;
......@@ -892,7 +891,7 @@ static inline int FUNC_NAME(pf_output_fp)(FUNC_NAME(puts_clbk) pf_puts, void *pu
ret += r;
f.Precision = three_digit_exp ? 3 : 2;
FUNC_NAME(pf_integer_conv)(buf, ARRAY_SIZE(buf), &f, radix_pos);
FUNC_NAME(pf_integer_conv)(buf, &f, radix_pos);
r = pf_puts(puts_ctx, f.Precision, buf);
if(r < 0) return r;
ret += r;
......@@ -1068,7 +1067,7 @@ int FUNC_NAME(pf_printf)(FUNC_NAME(puts_clbk) pf_puts, void *puts_ctx, const API
flags.PadZero = TRUE;
i = flags.Precision;
flags.Precision = 2*sizeof(void*);
FUNC_NAME(pf_integer_conv)(buf, ARRAY_SIZE(buf), &flags,
FUNC_NAME(pf_integer_conv)(buf, &flags,
(ULONG_PTR)pf_args(args_ctx, pos, VT_PTR, valist).get_ptr);
flags.PadZero = FALSE;
flags.Precision = i;
......@@ -1105,15 +1104,15 @@ int FUNC_NAME(pf_printf)(FUNC_NAME(puts_clbk) pf_puts, void *puts_ctx, const API
return -1;
if(flags.IntegerDouble || (flags.IntegerNative && sizeof(void*) == 8))
FUNC_NAME(pf_integer_conv)(tmp, max_len, &flags, pf_args(args_ctx, pos,
FUNC_NAME(pf_integer_conv)(tmp, &flags, pf_args(args_ctx, pos,
VT_I8, valist).get_longlong);
else if(flags.Format=='d' || flags.Format=='i')
FUNC_NAME(pf_integer_conv)(tmp, max_len, &flags,
FUNC_NAME(pf_integer_conv)(tmp, &flags,
flags.IntegerLength != LEN_SHORT ?
pf_args(args_ctx, pos, VT_INT, valist).get_int :
(short)pf_args(args_ctx, pos, VT_INT, valist).get_int);
else
FUNC_NAME(pf_integer_conv)(tmp, max_len, &flags,
FUNC_NAME(pf_integer_conv)(tmp, &flags,
flags.IntegerLength != LEN_SHORT ?
(unsigned)pf_args(args_ctx, pos, VT_INT, valist).get_int :
(unsigned short)pf_args(args_ctx, pos, VT_INT, valist).get_int);
......
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