Thanks for commenting about Windows, disk space etc. As and aside, I wonder if 
a check for sufficient disk space should be somewhere in  #snapshot:andQuit: 
, (for now I am just doing a "quit" on Linux only...)

Milan
On March 14, 2005 05:30 am, David T. Lewis wrote:
> I'm afraid that I do not know a way to do this on Windows. I don't know the
> right analog to unix signals for Win32, and I never put anything like this
> into OSProcess for Windows (I know that Win32 has a mechanism for passing
> console interrupts to a program, but I don't know how this should be done
> for a program running as a service, etc).
>
> As far as safety of the SIGHUP and SIGQUIT examples I suggested, I think
> that this is a safe mechanism. I have tried sending multiple SIQQUIT
> signals to Squeak from a shell script, and it still saves and exits
> correctly, so it should be fairly robust. I cannot test on OS X though (I
> am using Linux), so if you are running on that flavor of unix please test
> it carefully.
>
> One more thing I would suggest for a server: Add a check to make sure
> you have adequate disk space before saving the image. Otherwise, it
> would be possible to corrupt your image and changes files by trying to
> write them to a file system that was out of space. This is of course
> the case for saving a Squeak image anyway, but it seems like a good
> safety check if you are setting this up to automatically save the
> image in a server application.
>
> Also, you should of course run the Squeak server under some user ID such
> that only that user can kill the Squeak image by sending it a SIGQUIT or
> SIGHUP.
>
> Dave
>
> On Sun, Mar 13, 2005 at 08:20:48PM -0500, Milan Zimmermann wrote:
> > Dave, Thanks for your detail explanation - I am currently trying to
> > achieve "quit on stop", and sounds like sending the SIGHUP should work
> > without danger of loosing something (although perhaps me using your
> > example and do #quitPrimitive would be more correct). As well, is there
> > an equivalent way to register to "do this before quit" on Windows?
> >
> > Thanks Milan
> >
> > On March 13, 2005 10:39 am, David T. Lewis wrote:
> > > I think you can pick any signal number you want. Either SIGQUIT or
> > > SIGHUP sound like reasonable choices.
> > >
> > > If you want to use (for example) the SIGQUIT signal to stop Squeak with
> > > an image save, here is what to do:
> > >
> > > 1) Load OSProcess. You need this for the signal forwarding.
> > >
> > > 2) Evaluate the following in your Squeak image:
> > >
> > >   [OSProcess accessor forwardSigQuit wait.
> > >   SmalltalkImage current snapshot: true andQuit: true]
> > >     fork name: 'SIGQUIT watcher'
> > >
> > > 3) From a Unix shell, "$ kill -3 <squeakpid>". Squeak will save itself,
> > > then exit.
> > >
> > > Explanation: The #forwardSigQuit method registers a Semaphore as an
> > > external object, and arranges for the VM to forward SIGQUIT signals to
> > > the Squeak Semaphore. A process in Squeak waits on that Semaphore and
> > > does an image save and exit when the Semaphore receives a signal.
> > >
> > > Also, you can easily disable any other signal, as long as Unix allows
> > > that signal to be handled. For example, if you want SIGHUP to be
> > > ignored, just evaluate:
> > >
> > >   OSProcess accessor forwardSigHup
> > >
> > > This causes all Unix SIGHUP signals to be forwarded to another Squeak
> > > Semaphore registered as another external object. The signals are
> > > gobbled up by the Squeak Semaphore, so effectively they are just
> > > ignored.
> > >
> > > Dave
> > >
> > > On Sun, Mar 13, 2005 at 12:35:59AM -0500, Milan Zimmermann wrote:
> > > > Hi,
> > > >
> > > > A related question - is there a "correct" signal to "stop" Squeak
> > > > when run as "init.d" service? I am running a Kom HTTP Server in
> > > > headless mode (start and stop using the attached script). Currently I
> > > > use SIGHUP for "STOP". While it works, I wonder if there is a more
> > > > "correct" or "safe" (in the sense of not loosing the image or
> > > > whatever else could happen) signal to use...
> > > >
> > > > Thanks Milan
> > > >
> > > > On March 12, 2005 12:14 pm, Brent Pinkney wrote:
> > > > > Hi,
> > > > >
> > > > > I have a Smallwiki image rnning headless.
> > > > >
> > > > > I stop this image using: kill <pid> which Ipresume sends a SIGKILL
> > > > > signal to the image.
> > > > >
> > > > > Is there a way the image can trap a signal and execute some
> > > > > Smalltalk code, like saving the image ?
> > > > >
> > > > > Thanks
> > > > >
> > > > > Brent