Hello Alexander, Sergei, So I split it into two distinct patches. I've used the same idea in these two patches : For each field - stored the old value in record[1] (done by default for update) - evaluate fieId (in record[0]) - switch field pointer on record[1] Then restore all field pointer on record[0]. Main advantage of this solution : very small cpu overhead. It has only a memory footprint (need to allocate record[1] for m_var_table) Best regards, Jérôme.
-----Message d'origine----- De : Alexander Barkov [mailto:bar@mariadb.org] Envoyé : mercredi 27 septembre 2017 15:16 À : jerome brauge; 'Sergei Golubchik' Cc : MariaDB Developers (maria-developers@lists.launchpad.net) Objet : Re: MDEV-13417 / MDEV - 13418 The order of evaluation of SELECT..INTO assignments
Hi Jerome,
Can you please split the patch: extract and send the patch for MDEV-13417 first. Sergei will review it.
As for the MDEV-13418 part, I have an idea how to reuse a lot of code from MDEV-10591.
Thanks.
On 09/07/2017 06:32 PM, jerome brauge wrote:
Hello Sergei, Here is a patch for MDEV-13417 and MDEV-13418. Until you choose the right name for this new sql_mode, I temporarily chosen "SET_CONSISTENCY". It is set only for sql_mode=oracle for now.
Can you review it ? Thank you. Regard.