4:39 p.m.
Can someone verify I am not off my rocker here? Re :
https://hibernate.onjira.com/browse/HHH-7161
We have a case where Session defines a method that returns a public
interface org.hibernate.IdentifierLoadAccess. When SessionImpl
implements those methods it redefines the return to instead be a private
inner class that implements that org.hibernate.IdentifierLoadAccess
interface.
As I understood it, when accessed through the Session interface what
return type is defined by SessionImpl is irrelevant. And indeed I am
not able to reproduce this when access those methods via Session. My
IDE gives me errors if I try to cast that reference to SessionImpl
instead.
--
steve(a)hibernate.org
http://hibernate.org
5:35 p.m.
Hi,
I see it the same way as you do; narrowing down the return type in the
implementation class shouldn't be a problem.
I'm also getting no compilation error when compiling the user's code
(as you said, IdentifierLoadAccessImpl shouldn't even be visible at
the call site). Maybe it's a problem with a certain
compiler/environment only?
--Gunnar
2012/3/13 Steve Ebersole <steve(a)hibernate.org>:
Can someone verify I am not off my rocker here? Re :
https://hibernate.onjira.com/browse/HHH-7161
We have a case where Session defines a method that returns a public
interface org.hibernate.IdentifierLoadAccess. When SessionImpl
implements those methods it redefines the return to instead be a private
inner class that implements that org.hibernate.IdentifierLoadAccess
interface.
As I understood it, when accessed through the Session interface what
return type is defined by SessionImpl is irrelevant. And indeed I am
not able to reproduce this when access those methods via Session. My
IDE gives me errors if I try to cast that reference to SessionImpl
instead.
--
steve(a)hibernate.org
http://hibernate.org
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
6:55 p.m.
Hi Steve,
On Tue, Mar 13, 2012 at 10:39 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
As I understood it, when accessed through the Session interface what
return type is defined by SessionImpl is irrelevant. And indeed I am
not able to reproduce this when access those methods via Session. My
IDE gives me errors if I try to cast that reference to SessionImpl
instead.
I tested it quickly in the test case I was working on and I confirm it
works (with the Eclipse compiler and directly with Maven and the
Oracle JDK 1.6.30 compiler).
That said, I don't understand why you return the Impl instead of the
interface: it's not consistent with byNaturalId() and
bySimpleNaturalId() which are just below. Is there any reason why you
treated them differently?
--
Guillaume
10:46 a.m.
Hi Guillaume,
Was the initial problem you just using SessionImpl directly? Or is it a
problem of particular JDK handling the covariant return differently? In
the latter case we'd need to change the code. In the former, its a
discussion.
As for why byId is that way, TBH being new to being able to design APIs
with Java 5+ features I tend to try things to see how they work.
On 03/13/2012 06:55 PM, Guillaume Smet wrote:
Hi Steve,
On Tue, Mar 13, 2012 at 10:39 PM, Steve Ebersole<steve(a)hibernate.org> wrote:
> As I understood it, when accessed through the Session interface what
> return type is defined by SessionImpl is irrelevant. And indeed I am
> not able to reproduce this when access those methods via Session. My
> IDE gives me errors if I try to cast that reference to SessionImpl
> instead.
I tested it quickly in the test case I was working on and I confirm it
works (with the Eclipse compiler and directly with Maven and the
Oracle JDK 1.6.30 compiler).
That said, I don't understand why you return the Impl instead of the
interface: it's not consistent with byNaturalId() and
bySimpleNaturalId() which are just below. Is there any reason why you
treated them differently?
--
steve(a)hibernate.org
http://hibernate.org
11:51 a.m.
Hi Steve,
On Wed, Mar 14, 2012 at 4:46 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
Was the initial problem you just using SessionImpl directly? Or is
it a
problem of particular JDK handling the covariant return differently? In the
latter case we'd need to change the code. In the former, its a discussion.
I'm not the OP. It works perfectly on my side, as reported yesterday.
As for why byId is that way, TBH being new to being able to design
APIs with
Java 5+ features I tend to try things to see how they work.
The new APIs are really nice from a DSL point of view, especially the
natural id stuff. Thanks for your work on this.
IMHO, even if it's not a problem by itself, I think byId() should be
consistent with the rest of the API and return an interface. I don't
see any advantages to return the Impl in this case.
Have a nice day.
--
Guillaume
12:07 p.m.
Ah, that was Guenther that reported the problem... Sorry
On Wed 14 Mar 2012 11:51:30 AM CDT, Guillaume Smet wrote:
Hi Steve,
On Wed, Mar 14, 2012 at 4:46 PM, Steve Ebersole<steve(a)hibernate.org> wrote:
> Was the initial problem you just using SessionImpl directly? Or is it a
> problem of particular JDK handling the covariant return differently? In the
> latter case we'd need to change the code. In the former, its a discussion.
I'm not the OP. It works perfectly on my side, as reported yesterday.
> As for why byId is that way, TBH being new to being able to design APIs with
> Java 5+ features I tend to try things to see how they work.
The new APIs are really nice from a DSL point of view, especially the
natural id stuff. Thanks for your work on this.
IMHO, even if it's not a problem by itself, I think byId() should be
consistent with the rest of the API and return an interface. I don't
see any advantages to return the Impl in this case.
Have a nice day.
--
steve(a)hibernate.org
http://hibernate.org
2:37 a.m.
>Was the initial problem you just using SessionImpl directly?
Yes, that was the problem,
so probably there are no JDK's handling it differently.
regards
Günther
-----Original Message-----
From: hibernate-dev-bounces(a)lists.jboss.org [mailto:hibernate-dev-bounces@lists.jboss.org]
On Behalf Of Steve Ebersole
Sent: Wednesday, March 14, 2012 4:46 PM
To: Guillaume Smet
Cc: hibernate-dev(a)lists.jboss.org
Subject: Re: [hibernate-dev] Covariant returns
Hi Guillaume,
Was the initial problem you just using SessionImpl directly? Or is it a problem of
particular JDK handling the covariant return differently? In the latter case we'd
need to change the code. In the former, its a discussion.
As for why byId is that way, TBH being new to being able to design APIs with Java 5+
features I tend to try things to see how they work.
On 03/13/2012 06:55 PM, Guillaume Smet wrote:
Hi Steve,
On Tue, Mar 13, 2012 at 10:39 PM, Steve Ebersole<steve(a)hibernate.org> wrote:
> As I understood it, when accessed through the Session interface what
> return type is defined by SessionImpl is irrelevant. And indeed I am
> not able to reproduce this when access those methods via Session. My
> IDE gives me errors if I try to cast that reference to SessionImpl
> instead.
I tested it quickly in the test case I was working on and I confirm it
works (with the Eclipse compiler and directly with Maven and the
Oracle JDK 1.6.30 compiler).
That said, I don't understand why you return the Impl instead of the
interface: it's not consistent with byNaturalId() and
bySimpleNaturalId() which are just below. Is there any reason why you
treated them differently?
--
steve(a)hibernate.org
http://hibernate.org
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
4670
days inactive
4672
days old
6 comments
4 participants
participants (4)
-
Demetz, Guenther -
Guillaume Smet -
Gunnar Morling -
Steve Ebersole