On Wed, 16 Aug 2023 21:38:09 GMT, Stuart Marks <sma...@openjdk.org> wrote:
>> Thanks! Done > > @nikita-sakharin > > Thanks for finding this bug and offering to fix it! (And @shipilev thanks for > your assistance on this.) > > Putting the test into a separate JVM will work, but I don't think it's > necessary to actually allocate the space. The test is only testing the > indexes sent to `get` and `set` on the list, and it doesn't actually verify > the contents of the list. (Presumably that's done by other tests.) Therefore > it should be possible to create a "virtual" list of a given size that checks > that the indexes are all in bounds but that doesn't actually store any > elements. It should be fairly straightforward to do this by subclassing > AbstractList and overriding a few methods. > > The advantage of not actually allocating 4G of memory is that it makes it > easier to run a bunch of cases that test the boundary conditions. In fact I'd > like to see that in the test, as opposed to testing this one case. @stuart-marks Thanks for your review! I rewrote the test according with your advise. All checks passed successfully. Could you review the changes, please? ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/15270#discussion_r1297105328
