<Swing Dev> JDK 9 RFR of JDK-8043548: Fix raw and unchecked lint warnings in javax.swing.plaf.*

[Joe Darcy]

Hi Petr,

tl;dr -- I've made the changes you've suggested and they're reflected in

     http://cr.openjdk.java.net/~darcy/8043548.1/

I'll start the internal ccc paperwork now.

More comments below.

Thanks,

-Joe

On 06/24/2014 01:26 AM, Petr Pchelko wrote:
> Hello, Joe.
>
> Sorry for the delay. A couple of small comments:
>
> BasicOptionPaneUI:1235 - is the the SuppressWarning needed here?

No; upon double-checked the line

                 JList<?>  list = (JList)e.getSource();

will compile without warning; I've removed the SuppressWarnings annotation.


> BasicScrollBarUI:1269 - why are you using <Object> here? <?> would allow work the same and allow to avoid SuppressWarning

Yes; changed to just

                         JList<?> list = (JList)view;

without the annotation.

> BasicTableHeaderUI, BasicTableUI - you can replace Enumeration<?> to Enumeration<TableColumn> and avoid casts later. This applies to multiple places in these files.
> BasicTreeUI - you can replace Enumeration<?> with Enumeration<TreePath> and avoid several cats.

Good suggestion; changed.

> NimbusStyle:307 - do we need this SuppressWarning?

No; deleted.

> 		    :472 - you could reuse a helper function here too

Yeah; I wanted to keep the helper function private.

> 		    :863 - is this comment a leftover from your work? I don't think it's good for production, it's not very self-explanatory.

Yes; it is a leftover comment which I've now removed. (The State type 
has a bug in how it is generified; using raw types in this method is the 
least-bad way of accommodating State.)


> SynthComboBoxUI:223 - is this needed?

No; removed.

> SynthTableHeaderUI:244 - is this SuppressWarning really needed? It looks like after you've changed RowSorter to RowSorter<?> theres no warning on line 245 any more.

Removed.


>
> Thank you.
> With best regards. Petr.
>
> On 23 июня 2014 г., at 19:43, Joe Darcy <joe.darcy at oracle.com> wrote:
>
>> *ping*
>>
>> I'd like to push this change soon. (I know it is tedious to review; it was also tedious to generate ;-)
>>
>> Thanks,
>>
>> -Joe
>>
>> On 06/18/2014 05:01 PM, Joe Darcy wrote:
>>> Hello,
>>>
>>> Any comments on these proposed changes?
>>>
>>> Thanks,
>>>
>>> -Joe
>>>
>>> On 06/14/2014 06:13 PM, Joe Darcy wrote:
>>>> PS This change involves various API updates; once those are settled, I'll file the internal ccc paperwork.
>>>>
>>>> PPS The javax.swing.plaf.nimbus.State class is generified in an odd manner.
>>>>
>>>> -Joe
>>>>
>>>> On 06/14/2014 04:44 PM, Joe Darcy wrote:
>>>>> Hello,
>>>>>
>>>>> Please review the (unexpected large) fix for
>>>>>
>>>>>     JDK-8043548: Fix raw and unchecked lint warnings in javax.swing.plaf.*
>>>>>     http://cr.openjdk.java.net/~darcy/8043548.0/
>>>>>
>>>>> In most of the javax.swing.multi.MultiFoo.java type, I replaced a raw Vector with a Vector<ComponentUI> and was then able to delete a number of now redundant casts. In some of the MultiFoo classes,
>>>>>
>>>>> src/share/classes/javax/swing/plaf/multi/MultiPopupMenuUI.java
>>>>> src/share/classes/javax/swing/plaf/multi/MultiTabbedPaneUI.java
>>>>>     src/share/classes/javax/swing/plaf/multi/MultiTextUI.java
>>>>>     src/share/classes/javax/swing/plaf/multi/MultiTreeUI.java
>>>>>
>>>>> all the elements stored are actually a more specific type than ComponentUI, say, PopupMenuUI. In those cases, I still used a Vector<ComponentUI> rather than, say, a Vector<PopupMenuUI> to deal with other aspects of the API. In particular, the MultiFoo classes have a method
>>>>>
>>>>>     ComponentUI[] getUIs() methods
>>>>>
>>>>> which is computed over in MultiLookAndFeel. Generifying methods in MultiLookAndFeel, particularly creatUIs, to work properly in the face of generics would unfortunately be problematic.
>>>>>
>>>>> A few classes in javax.swing had to be updated now to fully clear the warnings in plaf; the rest of javax.swing will be handled in future work (JDK-8043550: Fix raw and unchecked lint warnings in javax.swing.*).
>>>>>
>>>>> Some additional changes were needed in sun/swing/BakedArrayList.java, but that class should probably be removed from the platform.
>>>>>
>>>>> I ran SwingSet2 with a build having these changes and nothing appeared obviously wrong.
>>>>>
>>>>> Thanks,
>>>>>
>>>>> -Joe