RFR: 8320796: CssMetaData.combine() [v6]

[Kevin Rushforth]

On Tue, 28 Nov 2023 23:34:30 GMT, Andy Goryachev <angorya at openjdk.org> wrote:

>> Provide a public utility method for use by custom and core Nodes to simplify boilerplate around styleable properties as well as to save some memory.
>> 
>> 
>>     /**
>>      * Utility method which combines {@code CssMetaData} items in one immutable list.
>>      * <p>
>>      * The intended usage is to combine the parent and the child CSS meta data for
>>      * the purposes of {@code getClassCssMetaData()} method, see for example {@link Node#getClassCssMetaData()}.
>>      * <p>
>>      * Example:
>>      * <pre>
>>      * private static final List<CssMetaData<? extends Styleable, ?>> STYLEABLES = CssMetaData.combine(
>>      *      <Parent>.getClassCssMetaData(),
>>      *      STYLEABLE1,
>>      *      STYLEABLE2
>>      *  );
>>      * </pre>
>>      * This method returns an instance of {@link java.util.RandomAccess} interface.
>>      *
>>      * @param list the css metadata items, usually from the parent, must not be null
>>      * @param items the additional items
>>      * @return the immutable list containing all of the items
>>      *
>>      * @since 22
>>      */
>>     public static List<CssMetaData<? extends Styleable, ?>> combine(
>>         List<CssMetaData<? extends Styleable, ?>> list,
>>         CssMetaData<? extends Styleable, ?>... items)
>> 
>> 
>> Problem(s):
>> 
>> - the same less-than-optimal implementation is duplicated throughout the javafx code base which combines the parent and child styleable properties, using ArrayList wrapped in a  Collections.unmodifiableList()
>> - the capacity of underlying ArrayList might exceed the number of items, wasting memory
>> - a potential exists for the custom component developer to inadvertently create a sub-standard implementation (i.e. return a List which does not implement RandomAccess interface), or forget to include parent's styleables.
>> 
>> Non-Goals:
>> 
>> - not redesigning the lazy initialization of STYLEABLES list
>> - not changing the way styleables are enumerated in Nodes and Controls
>> - not adding any new methods to JDK (collections)
>> 
>> To Be Discussed:
>> 
>> - it is not clear whether the order of styleables returned by <N extends Node>.getClassCssMetaData() is of any importance
>> - the current spec for Node.getCssMetaData() specifies the return value as "The CssMetaData associated with this node, which may include the CssMetaData of its superclasses.", implying that it may or may not include.  It is unclear whether it must include the parent's styleables (well, except the Node whose superclass is Obje...
>
> Andy Goryachev has updated the pull request incrementally with one additional commit since the last revision:
> 
>   review comments

OK, let's stick with `combine` then.

-------------

PR Comment: https://git.openjdk.org/jfx/pull/1296#issuecomment-1832598950