Re: ForkedProcess (Was: Kill and refresh)

[Andrew Cagney <cagney at redhat dot com>]

Mark Wielaard wrote:
> Hi,
>
> On Fri, 2006-07-14 at 14:49 +0200, Mark Wielaard wrote:
>   
> > On Fri, 2006-07-14 at 14:17 +0200, Mark Wielaard wrote:
> >     
> > > Yes, I actually have an addition to Fork which just gives you an
> > > Input/OutputStream to the process to communicate with. I found that a
> > > little more convenient then using signals. I'll post it to the list so
> > > you can see if that is something we might want to add.
> > >       
> > Here it is. I found it useful to write some quick tests which needed to
> > exchange information with the child process without needing to attach to
> > the process or send signals between frysk and the child process.
> >     
Two thoughts come to mind:

- we've a mechanism for communicating between frysk and a debug process 
using signals, rather than introducing custom mechanisms, can something 
based or following the existing technique be used?

- since it is only intended for testing and testing fork, can this live 
in TestLib - keeps frysk.sys's interfaces as close as possible to the 
underlying mechanism?

This code can't be plugged into frysk's event loop (I/O Stream can't be 
integrated into frysk.sys.Poll) so really isn't part of the core.

Andrew

> > 2006-07-14  Mark Wielaard  <mark@klomp.org>
> >
> >         * Fork.java (ForkedProcess): New static inner class.
> >         (ForkedInputStream): Likewise.
> >         (ForkedOutputStream): Likewise.
> >         (ForkedProcess): New static method.
> >         * cni/Fork.cxx (fork): New method.
> >         (ForkedOutputStream::write): Likewise.
> >         (ForkedInputStream::read(void)): Likewise.
> >         (ForkedInputStream::read(jbyteArray,jint,jint)): Likewise.
> >
> > At first I used Runtime.exec() for this, but since that spawns a new
> > Thread that tries to wait() on all processes that didn't work nicely
> > with the rest of frysk core.
> >
> > Would it make sense to add this new functionality to Fork?
> >     
>
> Since I am using this for writing new tests I would like it to be added.
> OK?
>
> Cheers,
>
> Mark
>
>