std: Add a nonblocking `Child::try_wait` method#38866

alexcrichton · 2017-01-06T06:52:11Z

This commit adds a new method to the Child type in the std::process module
called try_wait. This method is the same as wait except that it will not
block the calling thread and instead only attempt to collect the exit status. On
Unix this means that we call waitpid with the WNOHANG flag and on Windows it
just means that we pass a 0 timeout to WaitForSingleObject.

Currently it's possible to build this method out of tree, but it's unfortunately
tricky to do so. Specifically on Unix you essentially lose ownership of the pid
for the process once a call to waitpid has succeeded. Although Child tracks
this state internally to be resilient to multiple calls to wait or a kill
after a successful wait, if the child is waited on externally then the state
inside of Child is not updated. This means that external implementations of
this method must be extra careful to essentially not use a Child's methods
after a call to waitpid has succeeded (even in a nonblocking fashion).

By adding this functionality to the standard library it should help canonicalize
these external implementations and ensure they can continue to robustly reuse
the Child type from the standard library without worrying about pid ownership.

rust-highfive · 2017-01-06T06:52:15Z

r? @aturon

(rust_highfive has picked a reviewer for you, use r? to override)

alexcrichton · 2017-01-06T06:53:15Z

API-wise I was torn between returning io::Result<ExitStatus> and io::Result<Option<ExitStatus>>. I opted for the former as it's consistent with our "nonblocking" interfaces elsewhere, namely Read and Write for TCP streams and such.

aturon · 2017-01-06T22:59:49Z

@bors: r+

bors · 2017-01-06T22:59:50Z

📌 Commit 0048bf3 has been approved by aturon

aturon · 2017-01-06T23:00:09Z

@bors: r-

Sorry, needs to have a tracking issue first. But this LGTM.

This commit adds a new method to the `Child` type in the `std::process` module called `try_wait`. This method is the same as `wait` except that it will not block the calling thread and instead only attempt to collect the exit status. On Unix this means that we call `waitpid` with the `WNOHANG` flag and on Windows it just means that we pass a 0 timeout to `WaitForSingleObject`. Currently it's possible to build this method out of tree, but it's unfortunately tricky to do so. Specifically on Unix you essentially lose ownership of the pid for the process once a call to `waitpid` has succeeded. Although `Child` tracks this state internally to be resilient to multiple calls to `wait` or a `kill` after a successful wait, if the child is waited on externally then the state inside of `Child` is not updated. This means that external implementations of this method must be extra careful to essentially not use a `Child`'s methods after a call to `waitpid` has succeeded (even in a nonblocking fashion). By adding this functionality to the standard library it should help canonicalize these external implementations and ensure they can continue to robustly reuse the `Child` type from the standard library without worrying about pid ownership.

alexcrichton · 2017-01-07T05:20:49Z

@bors: r=aturon

bors · 2017-01-07T05:20:49Z

📌 Commit abb9189 has been approved by aturon

bors · 2017-01-08T21:48:09Z

⌛ Testing commit abb9189 with merge 7e7a00c...

bors · 2017-01-08T23:00:40Z

💔 Test failed - status-travis

alexcrichton · 2017-01-09T03:28:38Z

@bors: retry * osx linker segfaulted

…

On Sun, Jan 8, 2017 at 3:00 PM, bors ***@***.***> wrote: 💔 Test failed - status-travis <https://travis-ci.org/rust-lang/rust/builds/190079741> — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#38866 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAD95PDLjudA-GStT_K-nGYn7QU9IYFbks5rQWqbgaJpZM4LcdKN> .

bors · 2017-01-09T07:00:46Z

⌛ Testing commit abb9189 with merge 7aab3d3...

std: Add a nonblocking `Child::try_wait` method This commit adds a new method to the `Child` type in the `std::process` module called `try_wait`. This method is the same as `wait` except that it will not block the calling thread and instead only attempt to collect the exit status. On Unix this means that we call `waitpid` with the `WNOHANG` flag and on Windows it just means that we pass a 0 timeout to `WaitForSingleObject`. Currently it's possible to build this method out of tree, but it's unfortunately tricky to do so. Specifically on Unix you essentially lose ownership of the pid for the process once a call to `waitpid` has succeeded. Although `Child` tracks this state internally to be resilient to multiple calls to `wait` or a `kill` after a successful wait, if the child is waited on externally then the state inside of `Child` is not updated. This means that external implementations of this method must be extra careful to essentially not use a `Child`'s methods after a call to `waitpid` has succeeded (even in a nonblocking fashion). By adding this functionality to the standard library it should help canonicalize these external implementations and ensure they can continue to robustly reuse the `Child` type from the standard library without worrying about pid ownership.

bors · 2017-01-09T08:57:55Z

☀️ Test successful - status-appveyor, status-travis
Approved by: aturon
Pushing 7aab3d3 to master...

rust-highfive assigned aturon Jan 6, 2017

alexcrichton mentioned this pull request Jan 7, 2017

Tracking issue for Child::try_wait #38903

Closed

alexcrichton force-pushed the try-wait branch from 0048bf3 to abb9189 Compare January 7, 2017 05:20

bors merged commit abb9189 into rust-lang:master Jan 9, 2017

alexcrichton deleted the try-wait branch January 13, 2017 23:27

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

std: Add a nonblocking `Child::try_wait` method#38866