Fix for pg_upgrade's forcing pg_controldata into English

Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > Greg Sabino Mullane wrote:
> >> Specifically, LANGUAGE changes the headers of pg_controldata 
> >> (but not the actual output, LC_ALL does that). Thanks for the 
> >> nudge, I'll get to rewriting some code.
> 
> > pg_upgrade does this in controldata.c for this exact reason:
> 
> > 	    /*
> > 	     * Because we test the pg_resetxlog output strings, it has to be in
> > 	     * English.
> > 	     */
> > 	    if (getenv("LANG"))
> > 	        lang = pg_strdup(ctx, getenv("LANG"));
> > 	#ifndef WIN32
> > 	    putenv(pg_strdup(ctx, "LANG=C"));
> > 	#else
> > 	    SetEnvironmentVariableA("LANG", "C");
> > 	#endif
> 
> You do realize that's far from bulletproof?  To be sure that that does
> anything, you'd need to set (or unset) LC_ALL and LC_MESSAGES as well.
> And I thought Windows spelled it LANGUAGE not LANG ...

I have implemented your suggestion of setting LANG, LANGUAGE, and
LC_MESSAGES based on code in pg_regress.c;  that is the only place I see
where we force English, and it certainly has received more testing.

Should I set LC_ALL too?  The other variables mentioned in pg_regress.c?
Is this something I should patch to 9.0.X?

Patch attached.

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + It's impossible for everything to be true. +