PATCH: math i18n

PATCH: math i18n

[Clint Adams]

% export LC_ALL=pl_PL

% typeset -F a; ((a=1.1))
zsh: bad floating point constant

% ((a=1,1)) ; echo $a
1,0000000000

% LANG=C; ((a=1.1)) ; echo $a
1.1000000000

You can't make a floating point assignment if your locale's
decimal point is a comma.

The problem here is that strtod is using the locale yet zsh is
ignoring it.

This patch fixes the problem basically by allowing locale
to switch the meanings of '.' and ','.  Because I'm an
ignorant American, I have no idea what this breaks.

--- Src/math.c	1999/11/10 08:23:22	1.1.1.18
+++ Src/math.c	1999/11/10 17:03:30
@@ -184,9 +184,20 @@
 static int
 zzlex(void)
 {
+    char decimal = '.', thousands = ',';
     int cct = 0;
+#ifdef USE_LOCALE
+    struct lconv *lc;
+#endif
 
     yyval.type = MN_INTEGER;
+
+#ifdef USE_LOCALE
+    lc = localeconv();
+    decimal = *(lc->decimal_point);
+    thousands = *(lc->thousands_sep);
+#endif
+
     for (;; cct = 0)
 	switch (*ptr++) {
 	case '+':
@@ -324,7 +335,9 @@
 	case ':':
 	    return COLON;
 	case ',':
-	    return COMMA;
+	case '.':
+	    if (*(ptr-1) == thousands) return COMMA;
+	    else break;
 	case '\0':
 	    ptr--;
 	    return EOI;
@@ -349,15 +362,15 @@
 	    }
 	/* Fall through! */
 	default:
-	    if (idigit(*--ptr) || *ptr == '.') {
+	    if (idigit(*--ptr) || *ptr == decimal) {
 		char *nptr;
 		for (nptr = ptr; idigit(*nptr); nptr++);
 
-		if (*nptr == '.' || *nptr == 'e' || *nptr == 'E') {
+		if (*nptr == decimal || *nptr == 'e' || *nptr == 'E') {
 		    /* it's a float */
 		    yyval.type = MN_FLOAT;
 		    yyval.u.d = strtod(ptr, &nptr);
-		    if (ptr == nptr || *nptr == '.') {
+		    if (ptr == nptr || *nptr == decimal ) {
 			zerr("bad floating point constant", NULL, 0);
 			return EOI;
 		    }

Re: PATCH: math i18n

[Bart Schaefer]

On Nov 10, 12:11pm, Clint Adams wrote:
} Subject: PATCH: math i18n
}
} This patch fixes the problem basically by allowing locale
} to switch the meanings of '.' and ','.  Because I'm an
} ignorant American, I have no idea what this breaks.

It breaks the comma operator, which is the C-style sequential expression
evaluation operator.  That is,

	$((a=1,2))

should mean "assign 1 to `a' and return the value 2" and not "assign 1.2
to `a'".

I recommend that we either don't use strtod(), or find a way to force it
to use the C locale, rather than mess around with the shell syntax.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com