RFR: JDK-8199216: Memory leak and quadratic layout time with nested nodes (hbox) and pseudo-class in style sheet

[John Hendrikx]

On Thu, 30 Mar 2023 19:38:39 GMT, Michael Strauß <mstrauss at openjdk.org> wrote:

> Do we want to depend on this behavior? The specification of `Set.copyOf` says:
> 
> ```
> @implNote
> If the given Collection is an unmodifiable Set, calling copyOf will generally not create a copy.
> ```
> 
> While the actual implementation _never_ creates a copy if the given collection is immutable, the specification does not guarantee that in all cases, but only in the general case. This behavior could theoretically change without changing the specification of `Set.copyOf`.

If it ever changes implementation, I suppose we could do this logic ourselves -- however, you did give me an idea... before making the copy, I could check if it is in the cache already.  In the case that the set you pass is mutable, but I have an immutable copy already then no copy would need to be made.  The only time a copy is made then is if it wasn't in the cache yet.

Before arriving at this solution I had a class `ImmutablePseudoClassState` which extended `AbstractSet`.  But after stripping everything out that I didn't need, the only thing that was left was a single private field... which was also a `Set`.  I then concluded that there is no need for this wrapper at all, as everything in the CSS code can deal with `Set<PseudoClass>` directly.

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

PR Review Comment: https://git.openjdk.org/jfx/pull/1076#discussion_r1153751462