[PATCH/RFA]: Allow default locale different from "C"

[Corinna Vinschen]

Hi,

per POSIX, a system can have a default locale different from "C".
See http://www.opengroup.org/onlinepubs/7990989775/xbd/locale.html:

 "All implementations define a locale as the default locale, to be
  invoked when no environment variables are set, or set to the empty
  string. This default locale can be the POSIX locale or any other,
  implementation-dependent locale."

To allow this, the below patch adds a global variable which can be set
to some other value than the default value in a system-dependent manner.
On Cygwin we're planning to allow to set a default locale via a file,
like, for instance, /etc/default/locale, or /etc/sysconfig/locale.

Since we're heading straight to UTF-8 as default charset, this patch
also introduces a define DEFAULT_LOCALE, which is set to "C.UTF-8"
on Cygwin, to "C" otherwise.

Ok to apply?


Thanks,
Corinna


        * libc/locale/locale.c (DEFAULT_LOCALE): New define.
        (__default_locale): New global variable set to the default
        locale.
        (__get_locale_env): Return __default_locale rather than fixed
        string "C".


Index: libc/locale/locale.c
===================================================================
RCS file: /cvs/src/src/newlib/libc/locale/locale.c,v
retrieving revision 1.28
diff -u -p -r1.28 locale.c
--- libc/locale/locale.c 29 Sep 2009 19:12:28 -0000 1.28
+++ libc/locale/locale.c 7 Oct 2009 17:06:11 -0000
@@ -205,6 +205,20 @@ static char *categories[_LC_LAST] = {
 };
 
 /*
+ * Default locale per POSIX.
+ */
+#ifdef __CYGWIN__
+#define DEFAULT_LOCALE "C.UTF-8"
+#else
+#define DEFAULT_LOCALE "C"
+#endif
+/*
+ * This variable can be changed by any outside mechanism.  This allows,
+ * for instance, to load the default locale from a file.
+ */
+char __default_locale[ENCODING_LEN + 1] = DEFAULT_LOCALE;
+
+/*
  * Current locales for each category
  */
 static char current_categories[_LC_LAST][ENCODING_LEN + 1] = {
@@ -731,9 +745,9 @@ __get_locale_env(struct _reent *p, int c
   if (env == NULL || !*env)
     env = _getenv_r (p, "LANG");
 
-  /* 4. if none is set, fall to "C" */
+  /* 4. if none is set, fall to default locale */
   if (env == NULL || !*env)
-    env = "C";
+    env = __default_locale;
 
   return env;
 }


--
Corinna Vinschen
Cygwin Project Co-Leader
Red Hat

[Jeff Johnston]

On 07/10/09 01:07 PM, Corinna Vinschen wrote:

> Hi,
>
> per POSIX, a system can have a default locale different from "C".
> See http://www.opengroup.org/onlinepubs/7990989775/xbd/locale.html:
>
>   "All implementations define a locale as the default locale, to be
>    invoked when no environment variables are set, or set to the empty
>    string. This default locale can be the POSIX locale or any other,
>    implementation-dependent locale."
>
> To allow this, the below patch adds a global variable which can be set
> to some other value than the default value in a system-dependent manner.
> On Cygwin we're planning to allow to set a default locale via a file,
> like, for instance, /etc/default/locale, or /etc/sysconfig/locale.
>
> Since we're heading straight to UTF-8 as default charset, this patch
> also introduces a define DEFAULT_LOCALE, which is set to "C.UTF-8"
> on Cygwin, to "C" otherwise.
>
> Ok to apply?
>

I'm striving to remove all the OS/platform checks throughout shared code
so I would prefer you do it this way:

#ifndef DEFAULT_LOCALE
#define DEFAULT_LOCALE "C"
#endif

and then set DEFAULT_LOCALE in libc/include/sys/config.h or
configure.host for Cygwin.

That way, any platform may reset it if desired and we don't have a huge
if/else check in the code.

Ok to apply if you do it that way.

-- Jeff J.

>
> Thanks,
> Corinna
>
>
> * libc/locale/locale.c (DEFAULT_LOCALE): New define.
> (__default_locale): New global variable set to the default
> locale.
> (__get_locale_env): Return __default_locale rather than fixed
> string "C".
>
>
> Index: libc/locale/locale.c
> ===================================================================
> RCS file: /cvs/src/src/newlib/libc/locale/locale.c,v
> retrieving revision 1.28
> diff -u -p -r1.28 locale.c
> --- libc/locale/locale.c 29 Sep 2009 19:12:28 -0000 1.28
> +++ libc/locale/locale.c 7 Oct 2009 17:06:11 -0000
> @@ -205,6 +205,20 @@ static char *categories[_LC_LAST] = {
>   };
>
>   /*
> + * Default locale per POSIX.
> + */
> +#ifdef __CYGWIN__
> +#define DEFAULT_LOCALE "C.UTF-8"
> +#else
> +#define DEFAULT_LOCALE "C"
> +#endif
> +/*
> + * This variable can be changed by any outside mechanism.  This allows,
> + * for instance, to load the default locale from a file.
> + */
> +char __default_locale[ENCODING_LEN + 1] = DEFAULT_LOCALE;
> +
> +/*
>    * Current locales for each category
>    */
>   static char current_categories[_LC_LAST][ENCODING_LEN + 1] = {
> @@ -731,9 +745,9 @@ __get_locale_env(struct _reent *p, int c
>     if (env == NULL || !*env)
>       env = _getenv_r (p, "LANG");
>
> -  /* 4. if none is set, fall to "C" */
> +  /* 4. if none is set, fall to default locale */
>     if (env == NULL || !*env)
> -    env = "C";
> +    env = __default_locale;
>
>     return env;
>   }
>
>

[Corinna Vinschen]

On Oct  8 12:32, Jeff Johnston wrote:

> On 07/10/09 01:07 PM, Corinna Vinschen wrote:
>> Hi,
>>
>> per POSIX, a system can have a default locale different from "C".
>> See http://www.opengroup.org/onlinepubs/7990989775/xbd/locale.html:
>>[...]
>
> I'm striving to remove all the OS/platform checks throughout shared code so
> I would prefer you do it this way:
>
> #ifndef DEFAULT_LOCALE
> #define DEFAULT_LOCALE "C"
> #endif
>
> and then set DEFAULT_LOCALE in libc/include/sys/config.h or configure.host
> for Cygwin.
>
> That way, any platform may reset it if desired and we don't have a huge
> if/else check in the code.
>
> Ok to apply if you do it that way.

Thanks, applied.  I'm going to move the Cygwin-specific DEFAULT_LOCALE
into winsup/cygwin/include/cygwin/config.h.


Corinna

--
Corinna Vinschen
Cygwin Project Co-Leader
Red Hat