Daniel Fischer skrev:
> Magnus,
>
> Can you please review this patch? It's a fix for a non-critical
> portability issue on 64-bit big endian machines that would cause
> REPORT_PORT to be displayed as 0.
>
>> sql/set_var.cc@stripped, 2008-04-01 11:03:06+02:00,
>> df@stripped +1 -1
>> Unbreak REPORT_PORT on big endian machines where sizeof long !=
>> sizeof int.
>>
>> diff -Nrup a/sql/set_var.cc b/sql/set_var.cc
>> --- a/sql/set_var.cc 2008-03-31 09:40:36 +02:00
>> +++ b/sql/set_var.cc 2008-04-01 11:03:06 +02:00
>> @@ -713,7 +713,7 @@ static uchar *slave_get_report_port(THD return
>> (uchar*) &thd->sys_var_tmp.long_value;
>> }
>>
>> -static sys_var_readonly sys_repl_report_port(&vars, "report_port",
>> OPT_GLOBAL, SHOW_INT, slave_get_report_port);
>> +static sys_var_readonly sys_repl_report_port(&vars, "report_port",
>> OPT_GLOBAL, SHOW_LONG, slave_get_report_port);
>>
>> #endif
>
The patch looks fine, the value or report_port is stored in
"sys_var_tmp.long_value" which is returned as a char* from
'slave_get_report_port' and with the above read into a long and finally
displayed. ;)
But why didn't you file a bug report for this? Seems like there is a big
risk of this patch getting lost somewhere in the release tree otherwise.
/ Magnus
