<i18n dev> RFR: 8367703: Unneeded cast in java.text.DigitList.append

Roger Riggs rriggs at openjdk.org
Mon Sep 15 21:47:11 UTC 2025


On Mon, 15 Sep 2025 21:18:12 GMT, Justin Lu <jlu at openjdk.org> wrote:

> During parse routines, DecimalFormat uses DigitList to append digits from the parsed text.
> 
> Note that `digit` is always the int value 0 through 9 (and subsequently the code point 48 through 57) when passed to `append`.
> 
> Currently, `append` accepts a char which forces an int -> char -> byte conversion to be stored in `digits`. This can be simplified to int -> byte if the parameter type for the method is updated. The two call sites can safely make this swap.
> 
> Tiers 1-3 and java.text JCK tests continue to pass with this change.

Is there any observable/measureable performance difference? The byte type is a bit of a hack, while char is semantically more precise.

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

PR Comment: https://git.openjdk.org/jdk/pull/27300#issuecomment-3294063823


More information about the i18n-dev mailing list