Random speculation about the reason for PPC64 buildfarm failures:

maybe isalnum is returning a value with the low-order byte all zero?

diff --git a/src/backend/utils/adt/formatting.c b/src/backend/utils/adt/formatting.c
index 1ac3653..993e936 100644 (file)
--- a/src/backend/utils/adt/formatting.c
+++ b/src/backend/utils/adt/formatting.c
@@ -1,7 +1,7 @@
 /* -----------------------------------------------------------------------
  * formatting.c
  *
- * $PostgreSQL: pgsql/src/backend/utils/adt/formatting.c,v 1.148 2008/09/26 15:35:28 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/utils/adt/formatting.c,v 1.149 2008/10/06 05:03:27 tgl Exp $
  *
  *
  *      Portions Copyright (c) 1999-2008, PostgreSQL Global Development Group
@@ -1549,7 +1549,7 @@ char *
 str_initcap(const char *buff, size_t nbytes)
 {
        char            *result;
-       bool            wasalnum = false;
+       int                     wasalnum = false;
 
        if (!buff)
                return NULL;