Experimental Physics and
Industrial Control System

"J. Lewis Muir" <[email protected]> · Thu, 06 Feb 2014 18:33:00 -0600

On 2/6/14, 3:37 PM, Andrew Johnson wrote:
>> Was that a Windows IOC or a Cygwin IOC?
>
> That IOC was built for the cygwin-x86 target architecture.
>
> Hmm, trying it again on an IOC built using Microsoft VS10 shows the
> behaviour that Freddie described: stdout is line-buffered (I have to
> press return at the iocsh prompt for a background task's messages to
> appear), and stderr messages don't appear until I exit the IOC.
>
> Interesting, thanks for asking that.
>
> Your "test -t" check says that FDs 0, 1 and 2 are all connected to a
> terminal BTW.

Hi, Andrew.

OK, then maybe the problem is with Cygwin's pseudo terminal emulation
which uses pipes?  From what I can gather, this is a known weakness.
And a solution may not be straightforward since the Cygwin people would
have fixed it by now if it was?  I don't know enough about it to say.

The following mintty issue has some helpful comments (andy.koppe is the
mintty author):

  https://code.google.com/p/mintty/issues/detail?id=56

Comment #13 suggests the best way to avoid the problem is to use Cygwin
programs under Cygwin terminals and Windows programs under Windows
consoles.

Comment #49 mentions a program called winpty that adapts a Windows
console to the Cygwin terminal by creating a hidden Windows console and
polling it for changes.  It's at:

  https://github.com/rprichard/winpty

Lewis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System