Message 69374

Author
albertca
Date
2021-08-13.20:23:15
Message id
69374

Content

I'd advice to check performane in upcoming Postgres 14. It introduces some
optimizations which may make the patch unnecessary. The ones I remember
improved NOT IN queries but may be worth checking.

El dv., 13 d’ag. 2021, 13:44, Nicolas Évrard <bugs@tryton.org> va escriure:

>
> New submission from Nicolas Évrard <nicoe@b2ck.com>:
>
> We noticed that domains of the kind `['OR, ('a', '=', x), ('b.f1', '=',
> y)]` that result in an expression of the kind `t.a = x OR t.id in
> (subquery)` could be more optimal if we used an `UNION` of the two queries
> (because postgres can sometimes use an index but in my test even without
> indexes it's a bit more optimal).
>
> The review implements this on `Model.search`.
>
> ----------
> assignedto: nicoe
> keyword: review
> messages: 69366
> nosy: nicoe
> priority: feature
> reviews: 365781002
> status: in-progress
> title: Using indexes in queries where the filter is made of 'OR'-ed
> expressions is suboptimal
> type: performance
>
> ______________________________________
> Tryton issue tracker <bugs@tryton.org>
> <https://bugs.tryton.org/issue10658>
> ______________________________________
>
History
Date User Action Args
2021-08-13 20:23:15albertcasetrecipients: + nicoe, reviewbot
2021-08-13 20:23:15albertcalinkissue10658 messages
2021-08-13 20:23:15albertcacreate

Showing 10 items. Show all history (warning: this could be VERY long)