With the introduction of String.newStringWithLatin1Bytes, we can use it to
simplify the toString methods of Integer and Long. By replacing the
implementation that supports COMPACT_STRING = false with
newStringWithLatin1Bytes, we gain the following benefits:
1. Simplified implementation
2. Strings with codeSize < 35 in Integer.toString(int) and Long.toString(long)
can be inlined in C1, improving performance in early calls.
To make it easier to understand my proposal, I created a draft PR and hope to
get your feedback. https://github.com/openjdk/jdk/pull/27929
<https://github.com/openjdk/jdk/pull/27929 >
-
Shaojin Wen