OK thanks, Antony
In the meantime I think I had a breakthru, althought it might not be 'right'!
I worked out that I had to set the 'edges' aka graph->_table->in_use to
current_thd on entry to some methods and the crash seems to go away.
I eventually stumbled on that after quite lot of tracing code through the
guts of mariadb, and a red herring thinking we should be using a Handler_share
(we may well should still be, but that actualy seems to not be the cause of
the crash)
I dont actually know yet what any other ramifications of that might be... it
seems odd that MariaDB uses C++ but has public fields on classes that anthing
can set, I find that the intended interface is not really that well documented
and has to be guessed by looking at other code
--Andrew
On 31/10/14 18:18, Antony T Curtis wrote:
Sorry for the slow reply, I have been very busy at work. However I have not
been completely idle..
I have found my oqgraph+tokudb source.. It was unfortunate at the time that
MariaDB did not have tokudb and the only source download at that time was
simple tar.gz files so it will take me a bit more of my free time to figure
out what changes I had made.
Regards,
Antony
On Oct 23, 2014 5:08 AM, "Andrew McDonnell" <b...@andrewmcdonnell.net>
wrote:
Hi, bug MDEV-6282 I think is also rearing its head as MDEV-6345 and now as
https://mariadb.atlassian.net/browse/MDEV-6784 (please see my comment at
end)
Antony are you able to dig and find out what you did working with oqgraph
on toku?
--
Mailing list: https://launchpad.net/~oqgraph-dev
Post to : oqgraph-dev@lists.launchpad.net
Unsubscribe : https://launchpad.net/~oqgraph-dev
More help : https://help.launchpad.net/ListHelp
--
Mailing list: https://launchpad.net/~oqgraph-dev
Post to : oqgraph-dev@lists.launchpad.net
Unsubscribe : https://launchpad.net/~oqgraph-dev
More help : https://help.launchpad.net/ListHelp
