OK, So with the Bell milestone we will be supporting an upgrade path
(i.e. we will be concerned about backwards compatibility), so I think we
should *definitely* be testing not only the upgrade-ability of the table
protobuffer messages, but lots more things, including all the other
protobuffer messages defined for the kernel and APIs.
The method you describe below sounds like a sensible approach indeed.
However, can I make a small request and ask that the UDF
SHOW_TABLE_PROTO() be renamed to SHOW_TABLE_MESSAGE() (or
PRINT_TABLE_MESSAGE())? :)
As for using shell or Perl, sure though it should be easy enough to just
use the --exec test case command and/or SELECT INTO OUTFILE...
Cheers!
Jay
Stewart Smith wrote:
so, we should have testing for upgradability of the table proto.
An excellent simple task would be to take the show_table_proto() test
and get it to create all the tables, save the tables off into
tests/std_data, and then write a test that both does a SHOW CREATE
TABLE and SHOW_TABLE_PROTO() on them.
this should give us some level of assurance we haven't horribly broken
things in the future.
you should be able to do this with a bit of shell and/or perl.
possibly even using the original random query generator scripts there.
anyone game?
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
