Re: [psycopg] speed concerns with executemany()
От | Daniele Varrazzo |
---|---|
Тема | Re: [psycopg] speed concerns with executemany() |
Дата | |
Msg-id | CA+mi_8aFtNS3jN6OTn00kcLD8tGBcN61dMDxj3pAaP5xvG1Svg@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: [psycopg] speed concerns with executemany() (Christophe Pettus <xof@thebuild.com>) |
Ответы |
Re: [psycopg] speed concerns with executemany()
|
Список | psycopg |
On Mon, Jan 2, 2017 at 5:16 PM, Karsten Hilbert <Karsten.Hilbert@gmx.net> wrote: > On Mon, Jan 02, 2017 at 05:10:10PM +0100, Daniele Varrazzo wrote: > >> > Any chance 'l' can be supported to be a generator expression ? >> >> The function would consume only page_size records at time. This is to >> avoid creating a humongous query in memory and then on the server, but >> has the nice effect of requesting only as many records from the >> sequence. > > Nice. > > Can said sequence be a generator (IOW not needing to support > len() upfront) ? Correct: see the implementation of paginate() in the gist at <https://gist.github.com/dvarrazzo/4204cca5d1cd7c9c95bc814d81e6b83e>: the sequence is only consumed via next(). Final implementation may change but accessing the sequence only as a generator is a desired feature for me too. -- Daniele
В списке psycopg по дате отправления: