On Tue, 19 Aug 2025 09:18:06 GMT, Jaikiran Pai <j...@openjdk.org> wrote:
>> Marked as reviewed by dfuchs (Reviewer).
>
>> I see you removed the explicit mention of the FileChannel#read(ByteBuffer,
>> long) usage. I had placed it intentionally there, as requested by @dfuch,
>> for FileChannel can have custom implementations, and we better communicate
>> what FC method we use under the hood and with which assumptions.
>
> I felt that as long as we convey that the `FileChannel's` position isn't
> altered by the method, then it should be enough for the specification of the
> API, without having to explain which method the implementation uses. But I'll
> let Daniel suggest which one he prefers.
>
>> Regarding the _"the size of the {@code channel}"_ expression, maybe a
>> little bit grammar policing, but a channel doesn't have a size, instead, the
>> file referred by the channel has a size – `FC::size` Javadoc is worded in
>> this way too.
>
> I think it can be worded something like:
>
>> @throws IOException if the {@linkplain FileChannel#size() channel's size}
>> cannot be determined or the {@code channel} is closed
In 84d14fc0, improved the `ofFileChannel` Javadoc as suggested by @jaikiran –
did not include the mention of `FC::read` used under the hood.
-------------
PR Comment: https://git.openjdk.org/jdk/pull/26155#issuecomment-3201212175
