Re: closed STDOUT and open3

Replies are listed 'Best First'.
Re^2: closed STDOUT and open3 by ikegami (Patriarch) on Feb 05, 2010 at 00:13 UTC
Neither fork nor open3 does any ipc, so that's not true. open3 does setup the child's STD handles to do ipc (if so instructed), but that shouldn't depend on the parent's STD handles. I do consider this a bug in open3, but it's only triggered under odd, easily avoidable circumstances. ("It hurts when I do this." "So don't do that") — Yet. At the moment, it's not possible to distinguish between an exec failure and the child exiting with code 255, and it's not possible to capture the cause of the exec failure. I have a patch for open3 (waiting on Perl 5.12's release) that will have open3 communicate exec failures to the parent via ipc.	[reply]

Replies are listed 'Best First'.

Re^2: closed STDOUT and open3
by ikegami (Patriarch) on Feb 05, 2010 at 00:13 UTC

Neither fork nor open3 does any ipc*, so that's not true.

open3 does setup the child's STD handles to do ipc (if so instructed), but that shouldn't depend on the parent's STD handles.

I do consider this a bug in open3, but it's only triggered under odd, easily avoidable circumstances. ("It hurts when I do this." "So don't do that")

* — Yet. At the moment, it's not possible to distinguish between an exec failure and the child exiting with code 255, and it's not possible to capture the cause of the exec failure. I have a patch for open3 (waiting on Perl 5.12's release) that will have open3 communicate exec failures to the parent via ipc.

[reply]