Propose to remove support for CompactStrings off

[Alan Bateman]

On 25/10/2025 11:53, wenshao wrote:
> Initially, the option to disable CompactStrings was provided for 
> performance reasons.
>
> However, with some optimizations for UTF8 scenarios and Latin1, 
> disabling CompactStrings no longer has any performance benefits. I 
> propose removing the related code to make it simpler and easier to 
> maintain.
>

Thanks for bringing up this topic. I don't think it can be removed 
without notice, any removal would at least need to start with 
deprecating the CompactStrings VM option.

It would be useful to get list of the cases that dependent on 
-XX:-CompactString/equivalent. It was initially a contingency in the 
event of a performance or functional issue but I think we are past all 
that. Things to check are the 32-bit ARM port, is that this relevant? 
Other ports? IBM ports appear to care about -CompactString, maybe only 
with their own VM?

-Alan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/core-libs-dev/attachments/20251026/7cc8f68f/attachment-0001.htm>