Re: [boost] Subject: Formal Review of Proposed Boost.Process library starts tomorrow

Subject: Re: [boost] Subject: Formal Review of Proposed Boost.Process library starts tomorrow
From: Jeremy Maitin-Shepard (jeremy_at_[hidden])
Date: 2011-02-11 15:10:50

On 02/11/2011 03:14 AM, Oliver Kowalke wrote:
[snip]

> I'd like to see the possibility to send signals to the child process too.
>
> posix_child pchild( child);
> pchild.signal_sigstop();
> pchild.signal_sigcontinue();
> etc.

Sending signals to the child process is quite trivial: simply invoke the 
POSIX kill function with the pid of the child.  You don't even have to 
worry about looping to handle EINTR.  Potentially boost.process could 
provide an interface for this, but as it is specific to POSIX, and it is 
basically impossible to have a simpler interface than already provided 
by POSIX, it seems rather pointless.

In contrast, handling signals, and SIGCHLD in particular, is quite 
complicated, and therefore it would be quite useful for Boost.Process to 
provide proper handling, as already discussed.