[sqlalchemy] Re: backref VS relation on both sides
didnt get your attachment. but yes, they are different. during a
many-to-many flush() operation, the two relationships know to
communicate which one actually dealt with the row in the many-to-many
table, without it, youll get duplicate association rows.
additionally, the automatic management of the bi-directional
attribute in Python tends to smooth out operations significantly,
although two separate attributes should still work. but keep in mind
if you dont set up backref, you have to manually assign to each
collection in both directions. i dont know why youd want to do it
that way.
On Mar 27, 2007, at 7:40 AM, Gaetan de Menten wrote:
Hi list,
A while ago, someone came up with a problem on the Elixir list. The
problem is that in elixir we force users to define relationships on
both sides of the relationship (ie, there is no concept of backref).
And I (mistakenly) thought that having two mappers with relations to
each other was the same as having one mapper with a relation to the
other and a backref, so I didn't define backrefs for those
relationships.
It seems like, in fact, it's not exactly the same: the difference seem
to be with the automatic update of the other attribute (the backref)
*even before a flush is done*, as demonstrated at:
http://www.sqlalchemy.org/docs/
datamapping.html#datamapping_relations_backreferences
Now the question is: shouldn't these two ways to define relations
be equivalent?
Attached is a test case, demonstrating that the second way doesn't
work (in case it can be of any use).
BTW: there are several mistakes in the example there.
Address = mapper(Address, addresses_table)
User = mapper(User, users_table, properties = {
'addresses' : relation(Address, backref='user')
}
)
should read:
mapper(Address, addresses_table)
mapper(User, users_table, properties = {
'addresses' : relation(Address, backref='user')
}
)
and
a1.user is user and a2.user is user
True
should be:
a1.user is u and a2.user is u
True
--
Gaëtan de Menten
http://openhex.org
--~--~-~--~~~---~--~~
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
-~--~~~~--~~--~--~---
[sqlalchemy] Re: backref VS relation on both sides
On 3/27/07, Michael Bayer [EMAIL PROTECTED] wrote:
didnt get your attachment.
Yeah, I forgot it, as usual, but it wasn't very import anyway...
but yes, they are different. during a
many-to-many flush() operation, the two relationships know to
communicate which one actually dealt with the row in the many-to-many
table, without it, youll get duplicate association rows.
additionally, the automatic management of the bi-directional
attribute in Python tends to smooth out operations significantly,
although two separate attributes should still work.
It seems like my question wasn't clear, sorry about that, but your
answer contained what I wanted to know anyway. Just for the record (no
need to answer again), what I meant was: is it a bug that these two
ways behave differently or is it an intended behavior?.
Anyway, thanks for the clarification.
On Mar 27, 2007, at 7:40 AM, Gaetan de Menten wrote:
Hi list,
A while ago, someone came up with a problem on the Elixir list. The
problem is that in elixir we force users to define relationships on
both sides of the relationship (ie, there is no concept of backref).
And I (mistakenly) thought that having two mappers with relations to
each other was the same as having one mapper with a relation to the
other and a backref, so I didn't define backrefs for those
relationships.
It seems like, in fact, it's not exactly the same: the difference seem
to be with the automatic update of the other attribute (the backref)
*even before a flush is done*, as demonstrated at:
http://www.sqlalchemy.org/docs/
datamapping.html#datamapping_relations_backreferences
Now the question is: shouldn't these two ways to define relations
be equivalent?
Attached is a test case, demonstrating that the second way doesn't
work (in case it can be of any use).
BTW: there are several mistakes in the example there.
Address = mapper(Address, addresses_table)
User = mapper(User, users_table, properties = {
'addresses' : relation(Address, backref='user')
}
)
should read:
mapper(Address, addresses_table)
mapper(User, users_table, properties = {
'addresses' : relation(Address, backref='user')
}
)
and
a1.user is user and a2.user is user
True
should be:
a1.user is u and a2.user is u
True
--
Gaëtan de Menten
http://openhex.org
--
Gaëtan de Menten
http://openhex.org
--~--~-~--~~~---~--~~
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
-~--~~~~--~~--~--~---
