RFR: 8294696 - BufferedInputStream.transferTo should drain buffer when mark set [v2]

[Roman Kennke]

On Sat, 15 Oct 2022 13:17:27 GMT, Markus KARG <duke at openjdk.org> wrote:

> Isn't a test needed here which fails without but passes with the proposed change?

If you are saying, that a test should verify that, e.g. the wrapped OS receives the buffer upon calling BIS.transferTo() under certain circumstances, then no, I don't think such a test is useful, because that is an implementation detail. What should be tested is if BIS.transferTo() works correctly according to its contract, under all conceivable/possible/interesting conditions. If such a test already exists and exercises the changed code and verifies that it is still correct, then it should be ok IMO (and maybe mention in the PR which tests that would be), otherwise it would be useful to write and add such tests (either to this PR or to a preceding PR).

-------------

PR: https://git.openjdk.org/jdk/pull/10525