RFR: 8364361: [process] java.lang.Process should implement Closeable [v15]

[Roger Riggs]

On Wed, 8 Oct 2025 05:26:37 GMT, Alan Bateman <alanb at openjdk.org> wrote:

>> Roger Riggs has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains 22 additional commits since the last revision:
>> 
>>  - The ProcessCloseTest is modified to be able to test the logic
>>    around close() waiting for the process to exit and the specified
>>    behavior of interrupting the waitFor.
>>  - Close is modified to wait for the process to terminate before returning.
>>    Please review the revised javadoc.
>>    As suggested in review comments, waiting for the process to terminate
>>    allows a more orderly cleanup by the application.
>>    The streams are closed and close uses `waitFor()` with an unlimited timeout
>>    for the process to terminate.
>>    While waiting the thread can be `interrupted` to exit the `waitFor`.
>>    If/when `waitFor` is interrupted, the process is destroyedForcibly on
>>    all platforms. When close() returns, the thread interrupt will pending
>>    and can be handled by the caller.
>>    If a watchdog timeout is desired on close(), a separate thread
>>    can schedule an interrupt at a suitable time after close is called.
>>  - Review comment improvements to make expected behavior clearer.
>>  - Merge branch 'master' into 8364361-process-autocloseable
>>  - Clarify the use of waitFor after close() or T-W-R exit.
>>  - Additional review feedback.
>>    Updated to emphasize reading the data from the streams and calling waitFor
>>    to allow for normal termination.
>>  - Merge branch 'master' into 8364361-process-autocloseable
>>  - Address review comments for code and javadoc in Process, the test and the example.
>>    Reinforced that calling waitFor should occur before calling close to wait for termination and get the exitStatus.
>>    Corrected the error message check for invalid handles on Windows.
>>    Update test to allow normal completion when exceptions may be expected.
>>    Correct the expected message on Windows for an invalid handle exception.
>>  - Update close() to use "terminate" consistently.
>>  - Remove volatile from "closed" field; updates are guarded by synchronized.
>>  - ... and 12 more: https://git.openjdk.org/jdk/compare/e422ebfa...6f8f7327
>
> src/java.base/share/classes/java/lang/Process.java line 235:
> 
>> 233:                 destroyForcibly();
>> 234:                 // Re-assert the interrupt
>> 235:                 Thread.currentThread().interrupt();
> 
> I think this is still problematic as code after the t-w-r can't be guaranteed that the child has terminated. So I think it needs a loop that ends when waitFor completes without an exception (destroyForcibly would only be called on the first interrupt of course).

@AlanBateman can you comment on the possibility of looping indefinitely and not be interruptible if the OS process cannot be destroyed.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/26649#discussion_r2433581550