Hi folks,

Just wanted to check up on this to see if this issue is something which is
being tracked and if there was a potential fix in the works?

Thanks,
Lalit

On Thu, 2 Jan 2020 at 15:13, Lalit Maganti <lal...@google.com> wrote:

> Hi all,
>
> I believe that I have found a bug in the virtual table bytecode generation
> when OR constraints are present and argvIndex is set in xBestIndex but the
> application does not actually filter fully.
>
> The problem seems to be in not setting/unsetting the non-null flag
> correctly (SQLITE_JUMPIFNULL) on the instruction which does the comparison;
> this leads to NULLs slipping through the filter.
>
> To fully repro the problem, please find attached a small C file which
> shows the unexpected behaviour. By adding an EXPLAIN in-front of the SELECT
> query, you should be able to see the incorrect bytecode on the NE opcode.
>
> If the attachment doesn't come through for some reason, I've also put the
> same file as a GitHub gist
> <https://gist.github.com/tilal6991/57f8c51934379270abe58201c7b1dab1/>.
>
> Thank you!
>
> Regards,
> Lalit
>
_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users

Reply via email to