On Thu, 28 May 2026 05:58:55 GMT, Xueming Shen <[email protected]> wrote:
> 1. It appears the top-level Vector docs still describe the API as "six
> abstract subclasses of Vector corresponding to the supported set of element
> types ...". Should it be updated as part of this integration to mention
> Float16Vector as the newly added abstract subclass, with Float16 semantics
> and short carrier APIs, rather than leaving everything only in Float16Vector?
Updated Vector.java, other details on Float16Vector carrier and element type
differences are part of Float16Vector.java
> 2. VectorOperators.Conversion<E,F>
> domainType() and rangeType() still suggest E & F to be "a primitive type".
> With the new H conversions, the domain or range may be Float16.class, which
> is a supported lane type but not a Java primitive type. Should this be
> changed to “a supported lane type” or similar?
Hi @xuemingshen-oracle , Addressed both the feedbacks.
> src/jdk.incubator.vector/share/classes/jdk/incubator/vector/Float16Vector.java
> line 3150:
>
>> 3148: * for (int n = 0; n < ar.length; n++) {
>> 3149: * if (m.laneIsSet(n)) {
>> 3150: * ar[n] =
>> slice.getAtIndex(ValuaLayout.JAVA_SHORT.withByteAlignment(1), n);
>
> typo: Valu**e**Layout.JAVA_SHORT
These typo in mainline jdk also not related to this patch.
https://github.com/openjdk/jdk/blob/master/src/jdk.incubator.vector/share/classes/jdk/incubator/vector/ByteVector.java#L3430
https://github.com/openjdk/jdk/blob/master/src/jdk.incubator.vector/share/classes/jdk/incubator/vector/VectorShuffle.java#L334
Created a bug to address these: https://bugs.openjdk.org/browse/JDK-8385556
-------------
PR Comment: https://git.openjdk.org/jdk/pull/28002#issuecomment-4561704205
PR Review Comment: https://git.openjdk.org/jdk/pull/28002#discussion_r3316031561
