18 Sep
2012
18 Sep
'12
6:31 p.m.
On Tue, 18 Sep 2012 04:29:37 +0200, alfC <alfredo.correa@gmail.com> wrote:
[...]Thanks, the program worked (boost 1.48) and I posted the solution in stakoverflow: http://stackoverflow.com/a/12469478/225186 I don't have any idea why this works vs. the other. I wonder if I have to do the same for doing output to the process.
It's required to close the write-end of the pipe also in the parent process. The child process inherits the write-end and closes it. But it really closes only a copy. Once the file_descriptor_sink initialized with close_handle goes out of the scope, the write-end in the parent process is closed, too. HTH, Boris
[...]