On Tuesday 07 October 2008 00:01:58 Michael Bayer wrote:
> On Oct 6, 4:22 pm, [EMAIL PROTECTED] wrote:
> > hi
> > i have a query where
> > q = ...q.initial
> > q = q.join( some_link_name )
> > ...q.otherstuff
> > q = q.add_entity( target_klas_of_the_above_link)
> > ...
> > this used to work ok, but recently it went doing cartesian
> > products because the join made itself an alias.
>
> recently as of when ? q.join() will make a subquery of an alias
> if you're joining to a joined-table class. This because the syntax
> A join (X join Y) works very poorly, so it detects that and instead
> says A join (SELECT * from X join Y). I.e. treating the
> "selectable" of X join Y like the single unit it is. But 0.5 has
> been doing that since day one and 0.4 does it too, though 0.4 might
> not do it in the same set of cases.
>
> > i made it into:
> > q = ...q.initial
> > q = q.add_entity( target_klas_of_the_above_link)
> > q = q.join( some_link_name )
> > ...q.otherstuff
>
> that works ? does the join no longer create the subquery ?
neither makes a subquery. The child is declared as polymorphic
although it has no inheritants (with_polymorphic= ('*', None) ),
probably that triggers the aliasing... ahha i recently changed the
dbcook's select_table=... into with_polymorphic=... and that might be
the "when".
but anyway, it may break on another scenario where the child _is_
polymorphic.
here the situation:
q.join(child) then q.add-entity(child) does:
select child.whatever, parent.whatever
FROM child, X, parent join child as child_1
on parent.child_id == child_1.id LEFT OUTER JOIN
whatever-eagerloading
-> decartX
q.add-entity(child) then q.join(child) does:
select child_1.whatever, parent.whatever
FROM X, parent join child as child_1
on parent.child_id == child_1.id LEFT OUTER JOIN
whatever-eagerloading
-> no decartX, ok
so in either case the joined child is aliased, but in the working case
there is no another non-aliased in the FROM.
i tried the other way, query(child).join(parent-backref),
and this does not work either way:
q.add-entity(parent) then q.join(parent)
does add parent in the FROM, so decartX
q.join(parent) then q.add-entity(parent)
does not add parent in the FROM, but dies with:
no such column: parent.somecolumn
there is only parent_1 in the FROM, while the select wants
parent.all-the-columns.
to me there's something fishy about all this, it does not work (or
break) in the same way.
do u want it as ticket? not sure how to formulate it.
fixing dbcook to not give false with_polymorphics makes all cases work
as there no aliasing, but this is a workaround - IMO false
polymorphics should be ignored - or errored and this add_entity/join
business has to be fixed too.
ciao
svilen
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"sqlalchemy" group.
To post to this group, send email to sqlalchemy@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/sqlalchemy?hl=en
-~----------~----~----~----~------~----~------~--~---
