On Mon, 30 Mar 2026 16:58:28 GMT, Ruby Chen <[email protected]> wrote:
> When comparing a `MemorySegment` to `MemorySegment.NULL` using `==`, the
> comparison may fail because a segment with address `0` can be created
> independently and will therefore have a different identity from
> `MemorySegment.NULL`.
>
> This PR adds JavaDocs that explain this issue and warn against using `==` to
> compare segments to `MemorySegment.NULL`.
Great work! Left some comments.
src/java.base/share/classes/java/lang/foreign/MemorySegment.java line 1556:
> 1554: * the {@code NULL} segment is of 2<sup>62</sup>.
> 1555: *
> 1556: * @apiNote Clients should avoid using {@code ==} to compare a
> segment with
I think this is probably the only place where you need the comment. I don't
think adding this to `ofAddress` helps, as `ofAddress` is only one way to get a
segment with address `0`. There's other ways such as `reinterpret`, maybe on
some memory segment returned by a foreign call.
The main point we want to raise here is that NULL is not unique.
src/java.base/share/classes/java/lang/foreign/MemorySegment.java line 1558:
> 1556: * @apiNote Clients should avoid using {@code ==} to compare a
> segment with
> 1557: * {@code MemorySegment.NULL}. A segment with address
> {@code 0L} may be
> 1558: * created independently and may therefore have a
> different identity
"created independently" -- I would use `{@linkplain #ofAddress(long) created
independently}` so that the text also points to one of the problematic methods.
src/java.base/share/classes/java/lang/foreign/MemorySegment.java line 1559:
> 1557: * {@code MemorySegment.NULL}. A segment with address
> {@code 0L} may be
> 1558: * created independently and may therefore have a
> different identity
> 1559: * from {@code MemorySegment.NULL}. Furthermore, if {@code
> MemorySegment}
I think we can lose the para that starts with `Furthemore...`. When Valhalla
lands, we can revisit.
-------------
PR Review: https://git.openjdk.org/jdk/pull/30501#pullrequestreview-4035904798
PR Review Comment: https://git.openjdk.org/jdk/pull/30501#discussion_r3014578058
PR Review Comment: https://git.openjdk.org/jdk/pull/30501#discussion_r3014583356
PR Review Comment: https://git.openjdk.org/jdk/pull/30501#discussion_r3014586241
