Andrei Elkin wrote:
> Zhen Xing, hello.
>
> I am okay with the patch except one point, pls find inlined.
>
[snip]
> > === added file 'mysql-test/suite/binlog/t/binlog_write_error.test'
> > --- a/mysql-test/suite/binlog/t/binlog_write_error.test 1970-01-01 00:00:00
> +0000
> > +++ b/mysql-test/suite/binlog/t/binlog_write_error.test 2009-10-30 10:00:15
> +0000
> > @@ -0,0 +1,101 @@
> > +#
> > +# === Name ===
> > +#
> > +# binlog_write_error.test
> > +#
> > +# === Description ===
> > +#
>
> just english:
>
> > +# This test case check if the error of writing binlog file is proper
>
> properly
>
> > +# reported and handle when executing statements.
>
> handled
>
Will fix those.
[snip]
> >
> >
> > === modified file 'sql/sql_base.cc'
> > --- a/sql/sql_base.cc 2009-06-19 08:24:43 +0000
> > +++ b/sql/sql_base.cc 2009-10-30 10:00:15 +0000
> > @@ -1333,7 +1333,8 @@ void close_thread_tables(THD *thd)
> > handled either before writing a query log event (inside
> > binlog_query()) or when preparing a pending event.
> > */
> > - thd->binlog_flush_pending_rows_event(TRUE);
> > + /* TODO: I (hezx) think the error here should not be ignored */
> > + (void)thd->binlog_flush_pending_rows_event(TRUE);
>
>
> >> >> I don't think you need such TODO here. The error is not ignored
> but
> >> >> rather its check-out delayed till net_end_statement().
> >> >>
> >> >
> >> > I mean the code should check the error and return early.
> >> >
>
> In any case I suggest to convert this TODO into a new bug where btw you could
> describe not-addressed by this patch cases. We could discuss the issue on the bug
> page not
> here in the source code :-)
>
OK, I'll removed the todo and add a report on the bug.
> ...
>
> cheers,
>
> Andrei
>
