Stewart Smith wrote:
> yes, it's an innodb bug.
> not sure if one is already filed - can always file one though, worst
> case it'll be closed as a duplicate :)
It will be known as http://bugs.mysql.com/bug.php?id=20791 then.
> not your problem though, so don't worry too much (unless you want to
> submit a patch alongside it).
I bet you had a chance to look at the proposed code,
since you known what i did not broke :)
I don't know what the next step is (beside review to find what I did broke);
let me know if you need anything from me.
>> In other words, what kind of valgrind reports are considered worth
>> filing a bug / a patch ?
> all are worth filing bugs for. patches or suggestions are even better!
> Internally, we're becoming increasingly strict of no valgrind warnings
> in the tree, ever. However we're still cleaning some out.
Nice goal. It is the way to go.
I'll post reports when I bump into things then,
but only if I have some analysis and/or a proposed fix,
to save on the administrative part.