RFR: 8364361: [process] java.lang.Process should implement Closeable [v11]

[Jaikiran Pai]

On Mon, 29 Sep 2025 14:36:46 GMT, Roger Riggs <rriggs at openjdk.org> wrote:

>> src/java.base/share/classes/java/lang/Process.java line 186:
>> 
>>> 184:      * <p>
>>> 185:      * Before calling this method the caller should read the streams for any
>>> 186:      * data or text and call {@linkplain #waitFor() waitFor} if the exit value is needed.
>> 
>> Should we add some additional text which clearly highlights that waitFor() needs to be called (before close()) to allow the process to run to completion, even when the exit value or the stream contents aren't used/read by the calling application? I realize that a few lines above we specify:
>> 
>>> The streams are closed immediately and the process is terminated without waiting.
>> 
>> But maybe explicitly saying that waitFor() is necessary for allowing the process to run to completion would help reduce confusion? When I first saw the PR, my initial thought was `Process.close()` would wait for the process to run to completion and cleanup the resources that were used and I suspect it might become a common pattern where applications do something like:
>> 
>> 
>> try (Process p = new ProcessBuilder("some process that takes 5 seconds to complete normally").start()) {
>> }
>> 
>> and be surprised that the launched process was abruptly terminated. So maybe some clear text on this method would help?
>
> I think those warnings are already present.
> - `process is terminated without waiting`
> - `the caller should read the streams for any data or text and call waitFor if the exit value is needed`
> - `Calling waitFor before calling close or exiting the try-with-resources block allows the process time to clean up and exit.`
> - and the example calls waitFor inside the try-with-resources.
> 
> Does this add the emphasis you are asking for:
> `Calling waitFor after close() or after the try-with-resources block exits returns the status after destroying the process.`

I read the current proposed text a few more times. What you state is true - that it already covers the part that this method won't wait for the process to run to completion. However, I think, that detail doesn't register strongly when reading this text.

Here's what I had in mind - instead of the following sentence:


* Before calling this method the caller should read the streams for any
* data or text and call {@linkplain #waitFor() waitFor} if the exit value is needed.


would something like this be appropriate:


 * If the process should be allowed to run to completion, or the data from the
 * process stream(s) or the exit value of the process is of interest, then
 * the caller must {@linkplain #waitFor() wait} for the process to complete
 * before calling this method. 


(the rest of the text in the method doc can stay as-is) 

I think this expressly states that the applications must not call this `close()` method if they wish the process to run to completion.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/26649#discussion_r2391939947