Thanks for all your hard work tracking down this issue.
Jon.
On May 29, 8:02 pm, "Michael Bayer" <mike...@zzzcomputing.com> wrote:
> Michael Bayer wrote:
>
> > Michael Bayer wrote:
>
> >> Jonathan Marshall wrote:
>
> >>> How is the sqlalchemy's outptut of parameters generated? It has the
> >>> wrong value in it.
>
> >> hold that thought, I've reproduced the bug.
>
> > its a hash collision resulting from a hopefully undocumented usage of the
> > "params" dict you send to execute, whereby you can place a bindparam()
> > construct as a key in that dict. the bindparam() objects are created by
> > the text() construct, the hash values compare because after a lot of
> > values, you get a hash collision against the integer values being sent in,
> > and __eq__() returns True since all clauseelements produce an object
> > that's not False.
>
> > Since we had 0.5.4p3 scheduled for launch very soon, I'll look for a
> > workaround for this or remove the behavior altogether if I can get a sense
> > that nobody is using this feature.
>
> the behavior is removed in r5991. 0.5.4p3 is probably out today which
> was already planned.
>
> I'd be very surprised if anyone was using this behavior since it was never
> documented, it essentially looks like:
>
> bar = bindparam("bar")
> s = table.select().where(table.c.foo == bar)
> conn.execute(s, {bar:5})
>
> All of our documentation would write this as:
>
> conn.execute(s, {"bar":5})
>
> So its the {bar:5} feature that is removed.
--~--~---------~--~----~------------~-------~--~----~
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
sqlalchemy+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sqlalchemy?hl=en
-~----------~----~----~----~------~----~------~--~---
