Re: setlocale - Mailing list pgsql-hackers

From Peter Eisentraut
Subject Re: setlocale
Date
Msg-id Pine.LNX.4.44.0310081929570.26413-100000@peter.localdomain
Whole thread Raw
In response to setlocale  (Andrew Dunstan <andrew@dunslane.net>)
List pgsql-hackers
Andrew Dunstan writes:

>   At program startup, the equivalent of the following statement is executed:
>   setlocale( LC_ALL, "C" );
> Does this have any effect on us?

No, that is just a peculiar way to express that by default nothing
happens.

> Does it mean, as it appears to, that the locale will not be inherited
> from the parent?

A process never inherits the locale from the parent.  It only inherits
environment variables, among which may be LC_ALL, etc.  To activate any
kind of locale in a program you need to call

setlocale(LC_xxx, "something");

where "something" may be the name of the actual locale you want, or -- as
a special case -- it may be "", in which case it takes the value of the
respective environment variable LC_xxx.

There is an appearance of inheritance in shell scripts, but only because
the shell takes care of some of these things automatically.

> If so, I guess it could be got around by passing LC_COLLATE and LC_CTYPE
> arguments to postgres when running the bootstrap code.

The easiest solution would be to stick LC_COLLATE and LC_CTYPE into the
environment of the postgres process when you call it the first time (in
bootstrap mode).  If no --lc-* options were given, you don't need to do
anything, because "postgres" will just take what's in the environment.  If
--lc-* options where given, you could use putenv().

> Another question - will we want to internationalize initdb

Yes, but that should really wait until we have a working C version first.

-- 
Peter Eisentraut   peter_e@gmx.net



pgsql-hackers by date:

Previous
From: Andrew Dunstan
Date:
Subject: setlocale
Next
From: Neil Conway
Date:
Subject: confused about bit strings