[Dbix-class] patch for ResultSet::find_or_new

[Zbigniew Lukasiak]

On Jan 14, 2008 10:59 PM, Jason Kohles <email at jasonkohles.com> wrote:
>
> On Jan 14, 2008, at 2:19 PM, Zbigniew Lukasiak wrote:
>
> > On Jan 13, 2008 7:20 PM, Matt S Trout <dbix-class at trout.me.uk> wrote:
> >> On Sun, Jan 13, 2008 at 10:41:43AM +0100, Patrick Weemeeuw wrote:
> >>> There is a small bug in find_or_new: when the find part
> >>> fails, it calls new_result with the hash containing the
> >>> values that were used for the search. It should use no
> >>> values at all instead.
> >>
> >> This isn't a bug. If that's the behaviour you want, do
> >>
> >> my $o = $rs->find({ id => $id }) || $rs->new({});
> >>
> >>> Example buggy case: $o = $...->find_or_new( { id => $id } )
> >>> with id a not null primary key. When $id is undefined, there
> >>> is obviously no row in the DB, and a new result object is
> >>> returned. However, the object returned contains the column
> >>> id => NULL, which (1) is invalid for this kind of object,
> >>> and (2) prevents in some backends (e.g. Pg) that the
> >>> sequence is used to generate a unique id.
> >>
> >> So don't pass id if it isn't a valid value. Passing undef there is
> >> a bug
> >> in your code, not in DBIx::Class.
> >>
> >> The usual use of find_or_new is to pass a unique key plus additional
> >> attributes to be used for object creation (which are ignored in the
> >> find()
> >> by specifying the key attr as well). Consider for example
> >>
> >> my $stats = $schema->resultset('PageViews')->find_or_new(
> >>              { page => $page, views => 0 },
> >>              { key => 'page' }
> >>            );
> >>
> >
> > How can we excercise the _or_new part of find_or_new?
> >
>
> The results you are seeing are *exactly* the way it is expected to
> work.  In this example you are doing:
>
> $rs->find_or_new( { id => undef } )
>
> and getting back an object where id is undefined.  How is that
> mysterious or incorrect behaviour?  You told it 'find me a row in the
> database where id is undef, and if you can't find one, create a new
> object where id is undef' and that's exactly what it did.
>
> The big difference between ->find_or_new and ->find_or_create, is that
> ->find_or_new doesn't attempt to do the insert yet, and so is not
> guaranteed to return to you an object that even *can* be inserted.
> What I would do in this case is either not use ->find_or_new when you
> don't have a valid primary key, or do something like this:
>
> my $obj = $rs->find_or_new( $id ? { id => $id } : {} );
>
> or even...
>
> my $obj = $rs->find_or_new( { id => $id } );
> if ( ! defined $obj->id ) { $obj->id( 'foo' ) }
>
>
> >> From the above I conclude that we should omit 'page' in the request
> > (instead of setting it to 'undef'):
> >
>
> > my $stats = $schema->resultset('PageViews')->find_or_new(
> >               { views => 0 },
> >               { key => 'page' }
> >             );
> > but then if we have another record with views == 0 it will be found
> > and no new row would be created.
> >
> You've just created an even more bizarre situation, what you have now
> is essentially
>
> my $rs = $schema->resultset( 'PageViews' );
> my $stats = $rs->find( {} ) || $rs->new( { views => 0 } );
>
>
> The documentation for find_or_create says "Tries to find a record
> based on its primary key or unique constraint; if none is found,
> creates one and returns that instead."  What you are trying to do is
> use it without providing a primary key or a unique constraint, and the
> behavior you are seeing is exactly what I would expect to happen in
> that situation, since ->find can't match anything without a primary
> key or a unique constraint, you get back a new object instead.

No - that is not what would happen.  No new record will be created if
you have another record that have views == 0.
This is very important - the behaviour is not what is documented - but
it is intentional - as the comment in the code shows.

--
Zbigniew


> > This is because find falls back on finding by all columns in the query
> > (ie by 'views' in this case)
> > if the key is not represented in the query - this is not documented,
> > but it is commented in the source:
> >
> > # @unique_queries = () if there is no 'key' in $input_query
> >
> > # Build the final query: Default to the disjunction of the unique
> > queries,
> > # but allow the input query in case the ResultSet defines the query
> > or the
> > # user is abusing find
> > my $alias = exists $attrs->{alias} ? $attrs->{alias} : $self->{attrs}
> > {alias};
> > my $query = @unique_queries
> >    ? [ map { $self->_add_alias($_, $alias) } @unique_queries ]
> >    : $self->_add_alias($input_query, $alias);
> >
> > I would say - we should get rid of that 'special feature'.
> >
>
> I would say it's working exactly as intended, you are abusing find and
> it's doing the best it can to accomodate you, by creating new objects
> when your ->find fails to find a matching row.  The only alternative I
> can see would be for it to throw an exception when you try and call
> find_or_create without the
>
> --
> Jason Kohles, RHCA RHCDS RHCE
> email at jasonkohles.com - http://www.jasonkohles.com/
> "A witty saying proves nothing."  -- Voltaire
>
>
>
>
> _______________________________________________
> List: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/dbix-class
> IRC: irc.perl.org#dbix-class
> SVN: http://dev.catalyst.perl.org/repos/bast/DBIx-Class/
> Searchable Archive: http://www.grokbase.com/group/dbix-class@lists.rawmode.org
>



-- 
Zbigniew Lukasiak
http://brudnopis.blogspot.com/