pike.git/ src/ operators.c

2009-06-30

2009-06-30 14:14:57 by Martin Stjernholm <mast@lysator.liu.se>

• 48441941add48233df00e48b0eac4b6d725a2304 (15 lines) (+13/-2) ^{[ Show | Annotate ]}
  Branch: 7.9

Ensure that floats are printed with a decimal point. Fixes regression
introduced in rev 1.241.

Rev: src/operators.c:1.247

2:   || This file is part of Pike. For copyright information see COPYRIGHT.   || Pike is distributed under GPL, LGPL and MPL. See the file COPYING   || for more information. - || $Id: operators.c,v 1.246 2009/02/06 00:47:01 srb Exp $ + || $Id: operators.c,v 1.247 2009/06/30 14:14:57 mast Exp $   */      #include "global.h"
157:    "the NULL value":get_name_of_type(what->type),    ind->u.integer);    else if (ind->type == T_FLOAT) -  Pike_error ("Cannot index %s with %"PRINTPIKEFLOAT"g.\n", +  Pike_error ("Cannot index %s with %"PRINTPIKEFLOAT"e.\n",    (what->type == T_INT && !what->u.integer)?    "the NULL value":get_name_of_type(what->type),    ind->u.float_number);
483:    case T_FLOAT:    sprintf(buf,"%.*"PRINTPIKEFLOAT"g",    MAX_FLOAT_PREC_LEN, sp[-1].u.float_number); +  /* Ensure that a decimal point gets printed, since "g" removes it +  * completely if only zeroes would follow. Don't need to check for +  * an 'e', since if that is printed then "g" behaved like the "e" +  * format, and it always prints a '.' if the precision isn't +  * zero. */ +  if (!strchr (buf, '.')) +  strcat (buf, ".0");    break;       case T_INT:
1621:    case T_FLOAT:    sprintf(buffer,"%.*"PRINTPIKEFLOAT"g",    MAX_FLOAT_PREC_LEN, sp[e].u.float_number); +  /* See comment for T_FLOAT in o_cast_to_string. */ +  if (!strchr (buffer, '.')) +  strcat (buffer, ".0");   #ifdef PIKE_DEBUG    if (strlen (buffer) > MAX_FLOAT_SPRINTF_LEN)    Pike_fatal ("Formatted float %s is %"PRINTSIZET"u, "