Use HashMap#newHashMap, HashSet#newHashSet instead of CollectionUtil#newHashMap, CollectionUtil#newHashSet.

[vsop-479]

Description

Discussed in #14770.

[vsop-479]

Hello, @uschindler
There is a little difference between two versions on the calculation fo capacity: the old one divide by 0.75f, and add one. Now , we just divide 0.75f.

[uschindler]

Hi,
The plan was to remove both methods in main branch and replace the whole code by the default variant I n JDK.

The plus1 issue comes from a safety added, I think because @thecoop wanted to make sure that after rounding the size is correct. According to JDK docs the newHashXxx method in jdk ensures that the requested elements fit into the map without resizing. So they .ay gave changed other stuff to make the plus1 not needed.

Uwe

[thecoop]

Yup, these CollectionUtil methods can just be removed and the JCL methods used directly. There'll need to be a changelog entry documenting the API removal. CollectionUtil is marked as internal, so we can just remove these for Lucene 11

[vsop-479]

Thanks @uschindler, @thecoop.

I considered use the JDK methods directly, but chosed this way.

Anyway, I will fix it.

[vsop-479]

Hi @uschindler , @thecoop
I removed CollectionUtil#newHashMap, CollectionUtil#newHashSet, and use Use HashMap#newHashMap, HashSet#newHashSet directly.

This pr also plus 1 for CustomAnalyzer#paramsToMap 's map size.

[uschindler]

👍

[uschindler]

Merry Xmas! 🎄

[vsop-479]

Merry Xmas! 🎄

I will backport this to 10.x

[uschindler]

For 10.x maybe keep the CollectionUtil method as deprecated.