See it here on lines 323-352:
http://bazaar.launchpad.net/~bauble/bauble/trunk/annotate/head%3A/bauble/plugins/imex/csv_.py
--~--~-~--~~~---~--~~
You received this message because you are subscribed to the Google Groups
"sqlalchemy" group.
To post to this group,
I drastically sped up my inserts by precomputing any defaults on a
column and passing them explicitly instead of calculating them on each
insert. For example, each row had a timestamp and the timestamp was
being calculated on each insert for each row. Since I was inserting
them all at the same t
On Sun, Jan 18, 2009 at 1:46 PM, Ramiro Morales wrote:
>
> Take a look at ticket #6095 to know the motivation and rationale behind
> the addition of the through option and some of the advantages when
> compared with using you own model with FKs to the two related models.
>
> Also and the document
On Sun, Jan 18, 2009 at 1:32 PM, Victor Lin wrote:
>
> Hi,
>
> I am building some applications that insert data to database heavily.
> I commit in for-loop every iteration. It seems that it is not a smart
> way to commit data every iteration. SQL queries implies expensive IO
> action. So I think
inserting bulk data is most efficient if you use an insert many:
connection.execute(
table.insert(),
[{..row...}, {...row...}, {...row..} ..]
)
For additional performance, you want to ensure that the above insert
is not relying on any python-level default executions (like when
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 18.01.2009 16:32 Uhr, Victor Lin wrote:
> Hi,
> So I think that would be better to buffer data and flush them.
This is easily spoken the implementation pattern of SA: unit-of-work.
You might check the documentation.
- -aj
-BEGIN PGP SIGNATURE-
