Hi all,
I have to read thousands of rows from a table and compile some data, but
in certain conditions update those rows, all with same value. The ratio
of reads and writes here is widest possible. Sometimes no rows,
sometimes few, and sometimes all rows that are read have to be updated.
This depends upon the execution plan of the query and is more really a
postgresql question. Google postgresql IN performance and you will get
a good idea of it.
By the look of your code, Second option would obviously be faster as
it hits database once whereas first one flush after every change.
On Tue, Feb 21, 2012 at 3:24 PM, Manav Goel manav.goe...@gmail.com wrote:
This depends upon the execution plan of the query and is more really a
postgresql question. Google postgresql IN performance and you will get
a good idea of it.
By the look of your code, Second option would obviously be
Hi Mike,
I've been reading the Alembic docs, and the following popped out:
- The partial guid approach is nice for machines, but a nightmare for
humans, looking at a folder full of these can't be fun. Since it's the
link in the file that's important, could the files be given friendly
names
On Feb 21, 2012, at 12:57 PM, Chris Withers wrote:
Hi Mike,
I've been reading the Alembic docs, and the following popped out:
- The partial guid approach is nice for machines, but a nightmare for
humans, looking at a folder full of these can't be fun. Since it's the link
in the file
Thanks for your replies.
Using the IN list definitely speeds up the process, but I hate the
resulting query which uses bound variables for each and every element of
the list.
But I have another problem with this, there's a massive memory leak
somewhere. Take a look at this model:
class