On 4/19/10 11:44 AM, Kevin Lewis wrote:
> This is OK as long as the performance Schema files are still represented by a FRM
> file and 'performance_schema' directory on disk. This causes find_files to be called
> looking for a directory and FRM files. So as this relates to worklog 5308, I think this
> shows that it will be necessary to restructure find_files a little so that all virtual
> schemas can be recognized and the on-disk file search skipped for them.
Opening a table, using a FRM file or discovering it from a
handler::discover interface, happens later, and is not relevant here.
> The real purpose of your access checks for performance schema is to avoid anybody
> trying to write to that schema, correct? This change is consistent with that.
The purpose of this change is to report a proper error message when a
user attempts to select from a performance schema table.
Preventing writes to performance schema tables was already implemented,
and has not changed.
Thanks for the review.