variables test fails on Stewart's 64 bit SPARC machine

[Lee Bieber]

Complete log is at http://hudson.drizzle.org/view/Drizzle-build/job/drizzle-debian-sparc-64bit-build/2/console

This is build 1194 in the build branch

Stewart's machine:
% uname -a
Linux hurricane 2.6.30-1-sparc64-smp #1 SMP Sun Aug 16 21:46:34 UTC 2009 sparc GNU/Linux

main.variables [ fail ] --- ../tests/r/variables.result 2009-10-26 20:55:57.248608222 +0300
+++ ../tests/r/variables.reject 2009-10-26 21:48:29.416544727 +0300
@@ -524,14 +524,20 @@
 set @@global.error_count=1;
 ERROR HY000: Variable 'error_count' is a read only variable
 set @@max_heap_table_size= 4294967296;
+Warnings:
+Error 1292 Truncated incorrect max_heap_table_size value: '4294967296'
 select @@max_heap_table_size > 0;
 @@max_heap_table_size > 0
 1
 set global max_heap_table_size= 4294967296;
+Warnings:
+Error 1292 Truncated incorrect max_heap_table_size value: '4294967296'
 select @@max_heap_table_size > 0;
 @@max_heap_table_size > 0
 1
 set @@max_heap_table_size= 4294967296;
+Warnings:
+Error 1292 Truncated incorrect max_heap_table_size value: '4294967296'
 select @@max_heap_table_size > 0;
 @@max_heap_table_size > 0
 1

drizzletest: Result content mismatch

[fmpwizard (diego-fmpwizard)]

This is that lovely case where 32 bit machines return a warning, but 64 bit machines do not issue a warning, the final solution would be to just disable warnings on the test case, instead of the hacky way to detect 46/32bit machines I came up with.

I'll try to submit a branch for this bug later tonight.