On Fri, 23 Jan 2026 02:27:07 GMT, Vicente Romero <[email protected]> wrote:
>> some of the null checks javac generates are redundant and can be removed
>
> Vicente Romero has updated the pull request incrementally with one additional
> commit since the last revision:
>
> adding more tests
Looks a good start. We could stop here, or we could probably think a bit more
about how to propagate the info as to whether some sub-expression has already
been validated to be strict -- which will allow us to omit use site checks in
more cases.
src/jdk.compiler/share/classes/com/sun/tools/javac/comp/NullChecksWriter.java
line 271:
> 269: hasNonNullArgs(msym) &&
> 270: canBeOverriden) {
> 271: tree.args = newArgs(msym, tree.args);
The `newArgs` check should somehow be recursive -- e.g. one or more arguments
could be already known to be nulll restricted, so we don't really need a use
site check. E.g. there's many ingredients here:
1. Can we trust the signature of the method we're calling?
2. Can we trust the static nullness status of any of the method params?
An extra use site check should only be added if neither (1) or (2) hold
-------------
PR Review:
https://git.openjdk.org/valhalla/pull/1947#pullrequestreview-3696821451
PR Review Comment:
https://git.openjdk.org/valhalla/pull/1947#discussion_r2720606124
