Hello,

Do you think this behavior will be changed in NH3 RTM ? Actually, the
provider always generates an inner join (I think also it's a good idea
to generate an outer Join).

Cordially.

On 18 juil, 14:39, Fabio Maulo <fabioma...@gmail.com> wrote:
> About this matter we can talk a lot...
> Do you remember that times when to write hardcoded SQL everywhere was a bad
> choice ?
>
> Well... now I'm seeing LINQ-to-persistence wrote everywhere in the
> application and people continue saying me that is not the same.
>
> P.S. I have some white hair, a wife, a daughter and a dog and I know
> something about the immortality of the medusa.
>
>
>
>
>
> On Sun, Jul 18, 2010 at 9:20 AM, Frans Bouma <fr...@sd.nl> wrote:
> > > The main difference between LINQ and HQL is that HQL was designed to be
> > > translated to SQL.
> > > We have to design our systems to use the best query-tool for a specific
> > > task.
> > > I'm working with NH3 in production with more than one project using HQL,
> > > QueryOver, LINQ, H-SQL, SQL and even Criteria (variable no strongly typed
> > > queries).
> > > We don't have to limit our possibilities (EQO is with me ;) )
>
> >         It's also not wise to limit a query api to just Linq, as you can't
> > express all queries in linq, for example a simple join with an ON clause
> > using an expression other than an equality operator is not possible: you've
> > to revert to cross joins with a where clause, which not always can be
> > optimized properly by the RDBMS compared to an ansi join. This is IMHO als
> > one of the more cumbersome aspects of MS' O/R mappers and one which is
> > often
> > ignored by bookwriters/speakers, I wonder why ;)
>
> >                FB
>
> > > --
> > > Fabio Maulo
>
> > > El 18/07/2010, a las 07:48, Richard Birkby <rbir...@gmail.com> escribió:
>
> > > > On 18 Jul 2010, at 11:34, Frans Bouma <fr...@sd.nl> wrote:
>
> > > >> One could argue to always do an FKSide LEFT JOIN PKside, even if the
> > > >> fk side has no nullable fk fields, though it might be an RDBMS can
> > > >> optimize the INNER JOIN better than the LEFT JOIN.
>
> > > > Yes - SQL compact will not rearrange tables in a queryplan when it has
> > > > an outer join. Ironically, i've used this fact to force a more optimal
> > > > queryplan by using an outer join in HQL.
>
> > > > Richard
>
> --
> Fabio Maulo

Reply via email to