On Wed, 31 Mar 2010 17:37:26 -0000, Jay Pipes <email address hidden> wrote: > k, that gives me something to go on. thx!
If you want a simple way to know if you've fixed it, you can grab my embedded-innodb branch (specifically the auto-increment one) and use it.
just remove the transaction starting code from rnd-init and then run the auto_increment test in embedded_innodb suite.
(i'm pushing this up now)
InnoDB conviniently segfaults for you (with an error message!) if you try and use a NULL pointer as a transaction :)
on a side note, have we gotten rid of the need for ::store_lock to be starting a transaction yet? -- Stewart Smith
On Wed, 31 Mar 2010 17:37:26 -0000, Jay Pipes <email address hidden> wrote:
> k, that gives me something to go on. thx!
If you want a simple way to know if you've fixed it, you can grab my
embedded-innodb branch (specifically the auto-increment one) and use it.
just remove the transaction starting code from rnd-init and then run the
auto_increment test in embedded_innodb suite.
(i'm pushing this up now)
InnoDB conviniently segfaults for you (with an error message!) if you try
and use a NULL pointer as a transaction :)
on a side note, have we gotten rid of the need for ::store_lock to be
starting a transaction yet?
--
Stewart Smith