Partition by failure on postgres backend

[FedericoV]

I am working with Ibis version: '1.0.0+5.g80db366'

Trying to execute the following code:

seller_dimensions_table = monster.seller_dimension
last_snapshot_gb = seller_dimensions_table.groupby(['merchant_customer_id', 'snapshot_date'])
sorted_snapshot = last_snapshot_gb.order_by(seller_dimensions_table.snapshot_date).mutate(rank=ibis.row_number())
sorted_snapshot.execute(limit=10)

Gives this long traceback:

-----------------------------------------------------------------------
ProgrammingError                      Traceback (most recent call last)
~/anaconda3/lib/python3.6/site-packages/sqlalchemy/engine/base.py in _execute_context(self, dialect, constructor, statement, parameters, *args)
   1181                         parameters,
-> 1182                         context)
   1183         except BaseException as e:

~/anaconda3/lib/python3.6/site-packages/sqlalchemy/engine/default.py in do_execute(self, cursor, statement, parameters, context)
    469     def do_execute(self, cursor, statement, parameters, context=None):
--> 470         cursor.execute(statement, parameters)
    471 

ProgrammingError: Frame clause should not be specified for ranking window functions

The generated SQL looks like this:

SELECT ..., row_number() OVER (PARTITION BY t0.merchant_customer_id, t0.snapshot_date ORDER BY t0.snapshot_date ASC ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) - %(param_1)s AS rank 
FROM monster.seller_dimension AS t0

[cpcloud]

@FedericoV Thanks for the report. I'm going to release 1.1.0 in the next couple of weeks which will contain a fix for this issue.

[cpcloud]

@FedericoV What version of postgres are you using?

[FedericoV]

I'm using the postgres backend to work with Redshift, which usually works very well.

…

On Fri, Jun 7, 2019, 10:55 AM Phillip Cloud ***@***.***> wrote: @FedericoV <https://github.com/FedericoV> What version of postgres are you using? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1757?email_source=notifications&email_token=AAEZ25U2BNMBPVZ2CRDNNXLPZKOKJA5CNFSM4HFSGM4KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXGRQII#issuecomment-499980321>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEZ25XFN35LHIZYJLENJHDPZKOKJANCNFSM4HFSGM4A> .

[cpcloud]

Interesting, using postgres proper (not redshift) doesn't raise an error when specifying a frame clause for ranking functions. We should be able to remove them with no effect except to allow redshift to work.