RFC: 8357183: Improving efficiency of Writer::append(CharSequence) and Writer::append(CharSequence, int, int) / Sub Task of 8356679: Using CharSequence::getChars internally

[Alan Bateman]

On 24/05/2025 15:37, Markus KARG wrote:
>
> Chen,
>
>
> thank you for your approval for a DRAFT PR.
>
>
> Following your proposal, I have published a DRAFT PR at 
> https://github.com/openjdk/jdk/pull/25432 so everybody can see what 
> the actual intended code change looks like, and better focus on that 
> actual code change and its risks and benefits. The intention of that 
> DRAFT PR explicitly is NOT to request code reviews, but solely to 
> serve as a publicly visible piece of code to support the ongoing 
> discussion here on this mailing list.
>
>
> TARGET VERSION: 26
>
>
> As we are near to JDK 25 GA, the target is inclusion into JDK 26 or 
> later, so no need to rush. Nevertheless, I would be happy if everybody 
> comments ASAP (at least briefly) to keep this discussion alive.
>
>
> PRELIMINARY RISK ANALYSIS: LOW
>

Writer was retrofitted to implement Appendable in JDK 5 with the 3-arg 
append documented to behave "in exactly the same as ..." the 
write(String) method. That's 20 years of subclasses that might be 
depending on this behavior. Also 20 years of subclasses that may depend 
on the 3-arg append calling the 1-arg append. So my initial reaction is 
that the risk may not be low and this proposal will require significant 
analysis of Writer implementations before the compatibility risk can be 
assessed.

-Alan