Re: [Sqlalchemy-tickets] [sqlalchemy] #2746: Add a flag to explicitly enable correlation even when the subquery has `asfrom=True`

[sqlalchemy <mi...@zz...>]

#2746: Add a flag to explicitly enable correlation even when the subquery has
`asfrom=True`
------------------------------+----------------------------------
      Reporter:  sayap        |      Owner:  zzzeek
          Type:  enhancement  |     Status:  new
      Priority:  medium       |  Milestone:  0.9.0
     Component:  sql          |   Severity:  minor - half an hour
    Resolution:               |   Keywords:
Progress State:  in queue     |
------------------------------+----------------------------------
Changes (by zzzeek):

 * priority:  low => medium
 * status_field:  not decided upon => in queue
 * type:  task => enhancement
 * severity:  no triage selected yet => minor - half an hour
 * milestone:   => 0.9.0


Comment:

 well the funny thing here is that we've had a long-standing comment in the
 compiler, regarding the fact that the list of "Existing froms" we pass in
 is only from the immediate enclosing SELECT, not the SELECT that's
 enclosing that one.   I think at least we know we can take that out, since
 that only applies to auto-correlation and I think we can agree auto-
 correlation out of the FROM clause is not necessary.  This is the case
 where you say, "omit these elements from the FROM clause,
 unconditionally".

 I'd rather make this specific to each FROM for the "correlate" case.
 See the attached patch.

 {{{
 #!python
 t1 = table('t1', column('a'))
 t2 = table('t2', column('a'), column('b'))
 s = select(
     [t2.c.b],
     t1.c.a == t2.c.a,
 ).correlate_except(t2, froms=True).alias('s')
 s2 = select([func.foo(s.c.b)]).as_scalar()
 s3 = select([t1], order_by=s2)

 print s3

 }}}

--
Ticket URL: <http://www.sqlalchemy.org/trac/ticket/2746#comment:1>
sqlalchemy <http://www.sqlalchemy.org/>
The Database Toolkit for Python