RFR: 8304745: Lazily initialize byte[] in java.io.BufferedInputStream [v2]

[Eirik Bjorsnos]

On Thu, 23 Mar 2023 19:48:40 GMT, Eirik Bjorsnos <duke at openjdk.org> wrote:

>> Sergey Tsypanov has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Update src/java.base/share/classes/java/io/BufferedInputStream.java
>>   
>>   Co-authored-by: liach <7806504+liach at users.noreply.github.com>
>
> You are probably following #13121, but for everybody else's reference here is Alan's comment on that PR:
> 
> https://github.com/openjdk/jdk/pull/13121#issuecomment-1481127986

> @eirbjo you mean benchmark for BIS instantiation?

My general thought is that a benchmark demonstrating the  proposed performance benefits is always good and should perhaps be a requirement for such performance related changes. An effort should be made to detect regressions as well. This is just my personal preference, won't claim this is policy of any kind.

Until something is benchmarked, nobody knows if the benefit is just speculation.

I have experienced and suggested several PRs myself which benchmark work later showed dubious improvements which cased the change to be withdrawn. 

In #13121, I ran some (existing) benchmark on the PR only to discover it seems to have a weird, subtle regression.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/13150#issuecomment-1481900264