>>>>> "Tom" == Tom Holroyd <tomh@stripped> writes:
Tom> build errors:
Tom> 1) sql/sql_yacc.cc and sql/sql_acl.cc still need -O0 to work around bugs
Tom> in gcc 2.95.2.
The only fix for now is to compile with --with-low-memory
Tom> 2) The MYSQL_TYPE_ACCEPT macro in acinclude.m4 doesn't check for "unsigned
Tom> int", which is the correct answer on Linux/Alpha -- the compiler gives a
Tom> fatal error for "int".
I shall fix this ASAP.
Tom> runtime errors:
Tom> 1) select replace('aaaa', 'aa', 'b');
Tom> kills the server because of a bug in sql/sql_string.cc; patch follows:
Tom> --- sql_string.cc.orig Fri Nov 19 12:24:50 1999
Tom> +++ sql_string.cc Fri Nov 19 12:25:01 1999
Tom> @@ -355,7 +355,7 @@
Tom> bool String::replace(uint32 offset,uint32 arg_length,const String &to)
Tom> - long diff = (long) (to.length()-arg_length);
Tom> + long diff = ((long) to.length()-arg_length);
Tom> if (offset+arg_length <= str_length)
Tom> if (diff < 0)
Tom> This patch is necessary on the Alpha because longs are 8 bytes while ints
Tom> are 4, and when diff is (supposed to be) negative, the cast does not sign
Tom> extend the length difference because the length types are unsigned. So
Tom> when diff should be -1, it's set to 2^32 - 1 (a large positive long).
Tom> 2) load data local infile doesn't work. It just hangs and eventually gets
Tom> ERROR 1159: Got timeout reading communication packets
Tom> (same for mysqlimport --local). I don't know why.
Ouch; I have to wait until I get our Alpha (a couple of weeks) to
check this out...
PS: Sorry for the late reply; I am still trying to catch up with the
mails that piled up during my vacation.