On Tue, 5 Jan 2021 14:00:44 GMT, Chris Hegarty <che...@openjdk.org> wrote:
>> When the size of the memory map is zero, FileChannelImpl returns a `null`
>> Unmapper - this creates issues to the mapped memory segment implementation.
>>
>> To fix, I've created an empty mapped segment class which is initialized to
>> sensible defaults, and whose implenentation of force/load etc. do nothing.
>>
>> We already had a test for this condition - but the test was missing the
>> `@Test` annotation, so it was not run! I've now beefed up the test a bit to
>> make sure that mapped segment operations do not throw.
>
> src/jdk.incubator.foreign/share/classes/jdk/internal/foreign/MappedMemorySegmentImpl.java
> line 128:
>
>> 126: modes, scope);
>> 127: } else {
>> 128: return new EmptyMappedMemorySegmentImpl();
>
> While a little pedantic, should the `mapMode` be considered so that an empty
> mapped segment will feature the expected access modes when queried through
> the `accessModes` API ?
Ugh - yes, good point
-------------
PR: https://git.openjdk.java.net/jdk16/pull/83
