the bug #7499 is not more a bug, but my missunderstanding (RESOLVED INVALID)
My arguments are:
> create table tbl_test
> (id int not null primary key,
> str_last_name text not null,
> misc text);
> insert into tbl_test values
> (1, 'Kolesnik'),
> (83, 'GXXXXXXXXX'),
> (111, 'Kolesnik'),
> (175, 'GXXXXXXXXX');
> select id, str_last_name from tbl_test
> where id in (83,175,111,1) order by str_last_name;
> update tbl_test set misc = 'x' where id = 1;
> select id, str_last_name from tbl_test
> where id in (83,175,111,1) order by str_last_name;
> analyze tbl_test;
> select id, str_last_name from tbl_test
> where id in (83,175,111,1) order by str_last_name;
here you are right:
after "analyze tbl_test;"
the records with the str_last_name with value Kolesnik sorted now in
different order and
for the last name GXXXXXXXXX works the same.
>No, it asked to specify ORDER BY such that it "constrains the result
>rows into a unique order" -- which you are not doing in your
>examples. That is exactly what you *should* do to get the results
>you want.
here you are right also, because it seems now, that if "order by id" is missing
then data results of a query could vary depending on changes to a
record done(or other
algorythms).
Lets close this bug.
With deep respect,
Denis Kolesnik.