You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@wicket.apache.org by Pedro Santos <pe...@gmail.com> on 2017/03/26 23:39:07 UTC
Re: [wicket8] Use default method for IModel#detach()?
-0
I see no good reason for IModel to extend from IDetachable. Users should be
able to add this interface at their will.
>If IModel were a @FunctionalInterface, then you wouldn't need something like
a SupplierModel; you could just use a lambda directly:
IModel, as it's now, isn't a functional interface.
>Then maybe just make IModel#setObject() default too ..
There's no default implementation for IModel#setObject(). The one that was
added is as much semantically wrong as to say IModel is a functional
interface.
Proposal:
Why don't we make AbstractReadyOnlyModel the superinterface of IModel
instead of to keep it as an abstract adaptor for IModel? So we would have a
semantically correct functional interface.
Pedro Santos
On Tue, Oct 6, 2015 at 6:42 PM, Martin Grigorov <mg...@apache.org>
wrote:
> Ugh, right!
>
> Then maybe just make IModel#setObject() default too ...
> We will experiment! :)
>
> Martin Grigorov
> Wicket Training and Consulting
> https://twitter.com/mtgrigorov
>
> On Tue, Oct 6, 2015 at 11:27 PM, Michael Mosmann <mi...@mosmann.de>
> wrote:
>
> > .. @IReadOnlyModel: i hope, it will be easier to change then the last
> time
> > i tried it. Watch out for component default model stuff...
> >
> > :)
> >
> > Am 6. Oktober 2015 22:54:37 MESZ, schrieb Martin Grigorov <
> > mgrigorov@apache.org>:
> > >On Tue, Oct 6, 2015 at 10:43 PM, Andrew Geery <an...@gmail.com>
> > >wrote:
> > >
> > >> If IModel were a @FunctionalInterface, then you wouldn't need
> > >something
> > >> like a SupplierModel; you could just use a lambda directly:
> > >>
> > >> new Label("label", () -> "The current time is " + LocalDate.now());
> > >>
> > >
> > >This looks good indeed.
> > >It seems we will add IReadOnlyModel soon!
> > >
> > >
> > >>
> > >> And since IModel is Serializable, the lambda will be too, without
> > >having to
> > >> have an artificial interface that is both Serializable & a Supplier.
> > >>
> > >> Thanks
> > >> Andrew
> > >>
> > >> On Tue, Oct 6, 2015 at 4:21 PM, Martin Grigorov
> > ><mg...@apache.org>
> > >> wrote:
> > >>
> > >> > Same for IRequestHandler#detach()
> > >> >
> > >> > Martin Grigorov
> > >> > Wicket Training and Consulting
> > >> > https://twitter.com/mtgrigorov
> > >> >
> > >> > On Mon, Oct 5, 2015 at 10:42 PM, Martijn Dashorst <
> > >> > martijn.dashorst@gmail.com> wrote:
> > >> >
> > >> > > Should we use an empty default implementation for IModel#detach?
> > >> > >
> > >> > >
> > >> > > public class IModel<T> extends IDetachable
> > >> > > {
> > >> > > ...
> > >> > >
> > >> > > @Override
> > >> > > default void detach()
> > >> > > {
> > >> > > }
> > >> > > }
> > >> > >
> > >> > > This won't break existing applications, but might make it a bit
> > >easier
> > >> > > on the eyes to implement IModel directly.
> > >> > >
> > >> > > I'm not in favor of applying the default method to IDetachable,
> > >> > > because that would defeat the interface's purpose IMO.
> > >> > >
> > >> > > WDYT?
> > >> > >
> > >> > > Martijn
> > >> > >
> > >> >
> > >>
> >
> > --
> > Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail
> > gesendet.
>
Re: [wicket8] Use default method for IModel#detach()?
Posted by Pedro Santos <pe...@gmail.com>.
Hi Sven,
yes, I want to work on the proposals:
- to not force IDetachable to IModel
- IModel without the functional interface annotation
- named LambdaModel subclasses and a better name for this class
>IIRC Michael Mosmann tried to work on something similar but it never quite
worked out.
Indeed it looks a big change and I wouldn't propose it to Wicket 8, but
yes, I would like to give it a try in the future.
Pedro Santos
On Mon, Mar 27, 2017 at 1:50 PM, Sven Meier <sv...@meiers.net> wrote:
> Hi Pedro,
>
> >Why don't we make AbstractReadyOnlyModel the superinterface of IModel
>
> IIRC Michael Mosmann tried to work on something similar but it never quite
> worked out.
>
> Do you want to work on a proposal?
>
> Before you invest too much time into this: I'm pretty sure most devs won't
> be in favor of doing another round of IModel refactoring in Wicket 8
> because of semantics.
>
> Regards
> Sven
>
>
>
> On 27.03.2017 01:39, Pedro Santos wrote:
>
>> -0
>>
>> I see no good reason for IModel to extend from IDetachable. Users should
>> be
>> able to add this interface at their will.
>>
>> If IModel were a @FunctionalInterface, then you wouldn't need something
>>> like
>>>
>> a SupplierModel; you could just use a lambda directly:
>>
>> IModel, as it's now, isn't a functional interface.
>>
>> Then maybe just make IModel#setObject() default too ..
>>>
>> There's no default implementation for IModel#setObject(). The one that was
>> added is as much semantically wrong as to say IModel is a functional
>> interface.
>>
>> Proposal:
>>
>> Why don't we make AbstractReadyOnlyModel the superinterface of IModel
>> instead of to keep it as an abstract adaptor for IModel? So we would have
>> a
>> semantically correct functional interface.
>>
>>
>>
>> Pedro Santos
>>
>> On Tue, Oct 6, 2015 at 6:42 PM, Martin Grigorov <mg...@apache.org>
>> wrote:
>>
>> Ugh, right!
>>>
>>> Then maybe just make IModel#setObject() default too ...
>>> We will experiment! :)
>>>
>>> Martin Grigorov
>>> Wicket Training and Consulting
>>> https://twitter.com/mtgrigorov
>>>
>>> On Tue, Oct 6, 2015 at 11:27 PM, Michael Mosmann <mi...@mosmann.de>
>>> wrote:
>>>
>>> .. @IReadOnlyModel: i hope, it will be easier to change then the last
>>>>
>>> time
>>>
>>>> i tried it. Watch out for component default model stuff...
>>>>
>>>> :)
>>>>
>>>> Am 6. Oktober 2015 22:54:37 MESZ, schrieb Martin Grigorov <
>>>> mgrigorov@apache.org>:
>>>>
>>>>> On Tue, Oct 6, 2015 at 10:43 PM, Andrew Geery <an...@gmail.com>
>>>>> wrote:
>>>>>
>>>>> If IModel were a @FunctionalInterface, then you wouldn't need
>>>>>>
>>>>> something
>>>>>
>>>>>> like a SupplierModel; you could just use a lambda directly:
>>>>>>
>>>>>> new Label("label", () -> "The current time is " + LocalDate.now());
>>>>>>
>>>>>> This looks good indeed.
>>>>> It seems we will add IReadOnlyModel soon!
>>>>>
>>>>>
>>>>> And since IModel is Serializable, the lambda will be too, without
>>>>>>
>>>>> having to
>>>>>
>>>>>> have an artificial interface that is both Serializable & a Supplier.
>>>>>>
>>>>>> Thanks
>>>>>> Andrew
>>>>>>
>>>>>> On Tue, Oct 6, 2015 at 4:21 PM, Martin Grigorov
>>>>>>
>>>>> <mg...@apache.org>
>>>>>
>>>>>> wrote:
>>>>>>
>>>>>> Same for IRequestHandler#detach()
>>>>>>>
>>>>>>> Martin Grigorov
>>>>>>> Wicket Training and Consulting
>>>>>>> https://twitter.com/mtgrigorov
>>>>>>>
>>>>>>> On Mon, Oct 5, 2015 at 10:42 PM, Martijn Dashorst <
>>>>>>> martijn.dashorst@gmail.com> wrote:
>>>>>>>
>>>>>>> Should we use an empty default implementation for IModel#detach?
>>>>>>>>
>>>>>>>>
>>>>>>>> public class IModel<T> extends IDetachable
>>>>>>>> {
>>>>>>>> ...
>>>>>>>>
>>>>>>>> @Override
>>>>>>>> default void detach()
>>>>>>>> {
>>>>>>>> }
>>>>>>>> }
>>>>>>>>
>>>>>>>> This won't break existing applications, but might make it a bit
>>>>>>>>
>>>>>>> easier
>>>>>
>>>>>> on the eyes to implement IModel directly.
>>>>>>>>
>>>>>>>> I'm not in favor of applying the default method to IDetachable,
>>>>>>>> because that would defeat the interface's purpose IMO.
>>>>>>>>
>>>>>>>> WDYT?
>>>>>>>>
>>>>>>>> Martijn
>>>>>>>>
>>>>>>>> --
>>>> Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail
>>>> gesendet.
>>>>
>>>
>
Re: [wicket8] Use default method for IModel#detach()?
Posted by Sven Meier <sv...@meiers.net>.
Hi Pedro,
>Why don't we make AbstractReadyOnlyModel the superinterface of IModel
IIRC Michael Mosmann tried to work on something similar but it never
quite worked out.
Do you want to work on a proposal?
Before you invest too much time into this: I'm pretty sure most devs
won't be in favor of doing another round of IModel refactoring in Wicket
8 because of semantics.
Regards
Sven
On 27.03.2017 01:39, Pedro Santos wrote:
> -0
>
> I see no good reason for IModel to extend from IDetachable. Users should be
> able to add this interface at their will.
>
>> If IModel were a @FunctionalInterface, then you wouldn't need something like
> a SupplierModel; you could just use a lambda directly:
>
> IModel, as it's now, isn't a functional interface.
>
>> Then maybe just make IModel#setObject() default too ..
> There's no default implementation for IModel#setObject(). The one that was
> added is as much semantically wrong as to say IModel is a functional
> interface.
>
> Proposal:
>
> Why don't we make AbstractReadyOnlyModel the superinterface of IModel
> instead of to keep it as an abstract adaptor for IModel? So we would have a
> semantically correct functional interface.
>
>
>
> Pedro Santos
>
> On Tue, Oct 6, 2015 at 6:42 PM, Martin Grigorov <mg...@apache.org>
> wrote:
>
>> Ugh, right!
>>
>> Then maybe just make IModel#setObject() default too ...
>> We will experiment! :)
>>
>> Martin Grigorov
>> Wicket Training and Consulting
>> https://twitter.com/mtgrigorov
>>
>> On Tue, Oct 6, 2015 at 11:27 PM, Michael Mosmann <mi...@mosmann.de>
>> wrote:
>>
>>> .. @IReadOnlyModel: i hope, it will be easier to change then the last
>> time
>>> i tried it. Watch out for component default model stuff...
>>>
>>> :)
>>>
>>> Am 6. Oktober 2015 22:54:37 MESZ, schrieb Martin Grigorov <
>>> mgrigorov@apache.org>:
>>>> On Tue, Oct 6, 2015 at 10:43 PM, Andrew Geery <an...@gmail.com>
>>>> wrote:
>>>>
>>>>> If IModel were a @FunctionalInterface, then you wouldn't need
>>>> something
>>>>> like a SupplierModel; you could just use a lambda directly:
>>>>>
>>>>> new Label("label", () -> "The current time is " + LocalDate.now());
>>>>>
>>>> This looks good indeed.
>>>> It seems we will add IReadOnlyModel soon!
>>>>
>>>>
>>>>> And since IModel is Serializable, the lambda will be too, without
>>>> having to
>>>>> have an artificial interface that is both Serializable & a Supplier.
>>>>>
>>>>> Thanks
>>>>> Andrew
>>>>>
>>>>> On Tue, Oct 6, 2015 at 4:21 PM, Martin Grigorov
>>>> <mg...@apache.org>
>>>>> wrote:
>>>>>
>>>>>> Same for IRequestHandler#detach()
>>>>>>
>>>>>> Martin Grigorov
>>>>>> Wicket Training and Consulting
>>>>>> https://twitter.com/mtgrigorov
>>>>>>
>>>>>> On Mon, Oct 5, 2015 at 10:42 PM, Martijn Dashorst <
>>>>>> martijn.dashorst@gmail.com> wrote:
>>>>>>
>>>>>>> Should we use an empty default implementation for IModel#detach?
>>>>>>>
>>>>>>>
>>>>>>> public class IModel<T> extends IDetachable
>>>>>>> {
>>>>>>> ...
>>>>>>>
>>>>>>> @Override
>>>>>>> default void detach()
>>>>>>> {
>>>>>>> }
>>>>>>> }
>>>>>>>
>>>>>>> This won't break existing applications, but might make it a bit
>>>> easier
>>>>>>> on the eyes to implement IModel directly.
>>>>>>>
>>>>>>> I'm not in favor of applying the default method to IDetachable,
>>>>>>> because that would defeat the interface's purpose IMO.
>>>>>>>
>>>>>>> WDYT?
>>>>>>>
>>>>>>> Martijn
>>>>>>>
>>> --
>>> Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail
>>> gesendet.
Re: [wicket8] Use default method for IModel#detach()?
Posted by Pedro Santos <pe...@gmail.com>.
Neither I am sure I understand you. By dating the email thread and the
IModel interface hierarchy you mean that by them being old we can't discuss
their merit?
I meant that IModel isn't a functional interface, not that it isn't
annotated as one. Same for the setObject method not having a default
implementation.
IModel is the interface of objects with two behaviours, they can set and
get model objects. A functional interface is a contract promising only one
behavior.
How to throw an exception is the default functionality of setObject method?
Nothing is being set there, so it doesn't fit as a default implementation.
On Mar 27, 2017 3:22 AM, "Martin Grigorov" <mg...@apache.org> wrote:
Hi Pedro,
I am not sure I understand you!
This discussion was 1.5 years ago ...
On Mon, Mar 27, 2017 at 1:39 AM, Pedro Santos <pe...@gmail.com> wrote:
> -0
>
> I see no good reason for IModel to extend from IDetachable. Users should
be
> able to add this interface at their will.
>
IModel extends IDetachable since forever
https://github.com/apache/wicket/blob/wicket-1.3.x/jdk-
1.4/wicket/src/main/java/org/apache/wicket/model/IModel.java#L56
>
> >If IModel were a @FunctionalInterface, then you wouldn't need something
> like
> a SupplierModel; you could just use a lambda directly:
>
> IModel, as it's now, isn't a functional interface.
>
Yes, it is!
https://github.com/apache/wicket/blob/8a4e1b3c24f4ce1fc3e44ca1b5a923
0158d9b584/wicket-core/src/main/java/org/apache/wicket/model/IModel.java#L63
>
> >Then maybe just make IModel#setObject() default too ..
>
> There's no default implementation for IModel#setObject(). The one that was
>
Yes, there is!
https://github.com/apache/wicket/blob/8a4e1b3c24f4ce1fc3e44ca1b5a923
0158d9b584/wicket-core/src/main/java/org/apache/wicket/model/IModel.java#L81
> added is as much semantically wrong as to say IModel is a functional
> interface.
> Proposal:
>
> Why don't we make AbstractReadyOnlyModel the superinterface of IModel
> instead of to keep it as an abstract adaptor for IModel? So we would have
a
> semantically correct functional interface.
>
IMO the current design looks and works quite good! No need to change
anything!
>
>
>
> Pedro Santos
>
> On Tue, Oct 6, 2015 at 6:42 PM, Martin Grigorov <mg...@apache.org>
> wrote:
>
> > Ugh, right!
> >
> > Then maybe just make IModel#setObject() default too ...
> > We will experiment! :)
> >
> > Martin Grigorov
> > Wicket Training and Consulting
> > https://twitter.com/mtgrigorov
> >
> > On Tue, Oct 6, 2015 at 11:27 PM, Michael Mosmann <mi...@mosmann.de>
> > wrote:
> >
> > > .. @IReadOnlyModel: i hope, it will be easier to change then the last
> > time
> > > i tried it. Watch out for component default model stuff...
> > >
> > > :)
> > >
> > > Am 6. Oktober 2015 22:54:37 MESZ, schrieb Martin Grigorov <
> > > mgrigorov@apache.org>:
> > > >On Tue, Oct 6, 2015 at 10:43 PM, Andrew Geery <andrew.geery@gmail.com
> >
> > > >wrote:
> > > >
> > > >> If IModel were a @FunctionalInterface, then you wouldn't need
> > > >something
> > > >> like a SupplierModel; you could just use a lambda directly:
> > > >>
> > > >> new Label("label", () -> "The current time is " + LocalDate.now());
> > > >>
> > > >
> > > >This looks good indeed.
> > > >It seems we will add IReadOnlyModel soon!
> > > >
> > > >
> > > >>
> > > >> And since IModel is Serializable, the lambda will be too, without
> > > >having to
> > > >> have an artificial interface that is both Serializable & a
Supplier.
> > > >>
> > > >> Thanks
> > > >> Andrew
> > > >>
> > > >> On Tue, Oct 6, 2015 at 4:21 PM, Martin Grigorov
> > > ><mg...@apache.org>
> > > >> wrote:
> > > >>
> > > >> > Same for IRequestHandler#detach()
> > > >> >
> > > >> > Martin Grigorov
> > > >> > Wicket Training and Consulting
> > > >> > https://twitter.com/mtgrigorov
> > > >> >
> > > >> > On Mon, Oct 5, 2015 at 10:42 PM, Martijn Dashorst <
> > > >> > martijn.dashorst@gmail.com> wrote:
> > > >> >
> > > >> > > Should we use an empty default implementation for
IModel#detach?
> > > >> > >
> > > >> > >
> > > >> > > public class IModel<T> extends IDetachable
> > > >> > > {
> > > >> > > ...
> > > >> > >
> > > >> > > @Override
> > > >> > > default void detach()
> > > >> > > {
> > > >> > > }
> > > >> > > }
> > > >> > >
> > > >> > > This won't break existing applications, but might make it a bit
> > > >easier
> > > >> > > on the eyes to implement IModel directly.
> > > >> > >
> > > >> > > I'm not in favor of applying the default method to IDetachable,
> > > >> > > because that would defeat the interface's purpose IMO.
> > > >> > >
> > > >> > > WDYT?
> > > >> > >
> > > >> > > Martijn
> > > >> > >
> > > >> >
> > > >>
> > >
> > > --
> > > Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail
> > > gesendet.
> >
>
Re: [wicket8] Use default method for IModel#detach()?
Posted by Martin Grigorov <mg...@apache.org>.
Hi Pedro,
I am not sure I understand you!
This discussion was 1.5 years ago ...
On Mon, Mar 27, 2017 at 1:39 AM, Pedro Santos <pe...@gmail.com> wrote:
> -0
>
> I see no good reason for IModel to extend from IDetachable. Users should be
> able to add this interface at their will.
>
IModel extends IDetachable since forever
https://github.com/apache/wicket/blob/wicket-1.3.x/jdk-1.4/wicket/src/main/java/org/apache/wicket/model/IModel.java#L56
>
> >If IModel were a @FunctionalInterface, then you wouldn't need something
> like
> a SupplierModel; you could just use a lambda directly:
>
> IModel, as it's now, isn't a functional interface.
>
Yes, it is!
https://github.com/apache/wicket/blob/8a4e1b3c24f4ce1fc3e44ca1b5a9230158d9b584/wicket-core/src/main/java/org/apache/wicket/model/IModel.java#L63
>
> >Then maybe just make IModel#setObject() default too ..
>
> There's no default implementation for IModel#setObject(). The one that was
>
Yes, there is!
https://github.com/apache/wicket/blob/8a4e1b3c24f4ce1fc3e44ca1b5a9230158d9b584/wicket-core/src/main/java/org/apache/wicket/model/IModel.java#L81
> added is as much semantically wrong as to say IModel is a functional
> interface.
> Proposal:
>
> Why don't we make AbstractReadyOnlyModel the superinterface of IModel
> instead of to keep it as an abstract adaptor for IModel? So we would have a
> semantically correct functional interface.
>
IMO the current design looks and works quite good! No need to change
anything!
>
>
>
> Pedro Santos
>
> On Tue, Oct 6, 2015 at 6:42 PM, Martin Grigorov <mg...@apache.org>
> wrote:
>
> > Ugh, right!
> >
> > Then maybe just make IModel#setObject() default too ...
> > We will experiment! :)
> >
> > Martin Grigorov
> > Wicket Training and Consulting
> > https://twitter.com/mtgrigorov
> >
> > On Tue, Oct 6, 2015 at 11:27 PM, Michael Mosmann <mi...@mosmann.de>
> > wrote:
> >
> > > .. @IReadOnlyModel: i hope, it will be easier to change then the last
> > time
> > > i tried it. Watch out for component default model stuff...
> > >
> > > :)
> > >
> > > Am 6. Oktober 2015 22:54:37 MESZ, schrieb Martin Grigorov <
> > > mgrigorov@apache.org>:
> > > >On Tue, Oct 6, 2015 at 10:43 PM, Andrew Geery <andrew.geery@gmail.com
> >
> > > >wrote:
> > > >
> > > >> If IModel were a @FunctionalInterface, then you wouldn't need
> > > >something
> > > >> like a SupplierModel; you could just use a lambda directly:
> > > >>
> > > >> new Label("label", () -> "The current time is " + LocalDate.now());
> > > >>
> > > >
> > > >This looks good indeed.
> > > >It seems we will add IReadOnlyModel soon!
> > > >
> > > >
> > > >>
> > > >> And since IModel is Serializable, the lambda will be too, without
> > > >having to
> > > >> have an artificial interface that is both Serializable & a Supplier.
> > > >>
> > > >> Thanks
> > > >> Andrew
> > > >>
> > > >> On Tue, Oct 6, 2015 at 4:21 PM, Martin Grigorov
> > > ><mg...@apache.org>
> > > >> wrote:
> > > >>
> > > >> > Same for IRequestHandler#detach()
> > > >> >
> > > >> > Martin Grigorov
> > > >> > Wicket Training and Consulting
> > > >> > https://twitter.com/mtgrigorov
> > > >> >
> > > >> > On Mon, Oct 5, 2015 at 10:42 PM, Martijn Dashorst <
> > > >> > martijn.dashorst@gmail.com> wrote:
> > > >> >
> > > >> > > Should we use an empty default implementation for IModel#detach?
> > > >> > >
> > > >> > >
> > > >> > > public class IModel<T> extends IDetachable
> > > >> > > {
> > > >> > > ...
> > > >> > >
> > > >> > > @Override
> > > >> > > default void detach()
> > > >> > > {
> > > >> > > }
> > > >> > > }
> > > >> > >
> > > >> > > This won't break existing applications, but might make it a bit
> > > >easier
> > > >> > > on the eyes to implement IModel directly.
> > > >> > >
> > > >> > > I'm not in favor of applying the default method to IDetachable,
> > > >> > > because that would defeat the interface's purpose IMO.
> > > >> > >
> > > >> > > WDYT?
> > > >> > >
> > > >> > > Martijn
> > > >> > >
> > > >> >
> > > >>
> > >
> > > --
> > > Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail
> > > gesendet.
> >
>