Horton

Re: Per-column collation, the finale

Noah Misch <noah@leadboat.com>

View thread
From: Noah Misch <noah@leadboat.com>
To: Peter Eisentraut <peter_e@gmx.net>, Itagaki Takahiro <itagaki.takahiro@gmail.com>
Cc: pgsql-hackers@postgresql.org
Date: 2011-01-26T03:32:24Z
Lists: pgsql-hackers 
On Wed, Jan 26, 2011 at 12:35:07AM +0200, Peter Eisentraut wrote:
> On tis, 2011-01-25 at 10:14 +0900, Itagaki Takahiro wrote:
> > and I have an almost empty pg_collation catalog with it:
> > 
> > =# SELECT * FROM pg_collation;
> >  collname | collnamespace | collencoding | collcollate | collctype
> > ----------+---------------+--------------+-------------+-----------
> >  default  |            11 |            0 |             |
> > (1 row)
> 
> The initdb output should say something about how it got there.

FWIW, I tried and had the same problem.  initdb gave:

  creating collations ... not supported on this platform

"configure" was failing to detect locale_t for me, and this fixed it:

*** a/configure.in
--- b/configure.in
***************
*** 1116,1122 **** AC_TYPE_INTPTR_T
  AC_TYPE_UINTPTR_T
  AC_TYPE_LONG_LONG_INT
  
! AC_CHECK_TYPES([locale_t],
  [#include <locale.h>])
  
  AC_CHECK_TYPES([struct cmsgcred, struct fcred, struct sockcred], [], [],
--- 1116,1122 ----
  AC_TYPE_UINTPTR_T
  AC_TYPE_LONG_LONG_INT
  
! AC_CHECK_TYPES([locale_t], [], [],
  [#include <locale.h>])
  
  AC_CHECK_TYPES([struct cmsgcred, struct fcred, struct sockcred], [], [],