I've written up a draft JEP for deprecating the disabling of Compact Strings.
There hasn't been a good term for running the system with Compact Strings
disabled, so I made up a term "UTF-16-only" and used it here.
https://openjdk.org/jeps/8371379
This is deprecating something that I assume most developers have never heard of.
It's important that anyone reading the JEP will quickly see that this is not
deprecating String, it's not dropping UTF-16, and it's not changing the String
constructor that take a Charset. Instead it's all about JDK internal representation
of String. No change needed to my code. I think the JEP needs a clear call-to-
action: Audit your application launch scripts and configuration to see if you are
using -XX:-CompactStrings. If you are, remove it, and re-test/re-benchmark to see if
see if there is any impact, preferably with a recent JDK release.
Good points about what the JEP should say. I've updated it.
Pretty much every article I've seen that describes Compact Strings also mentions the
-XX:-CompactStrings option, so the option seems fairly well known. (In addition to
it appearing in our documentation.) However, we don't know how many people actually
*use* the option. I suspect there might be a long tail of conservative ("don't touch
anything!") folks who added it a long time ago out of an abundance of caution and
forgot about it, so it would be good to get the word out as widely as possible.
s'marks
