OK. But why do you find that just killing the hung client and restarting it
via the NSM GUI is not sufficient to restore your session's function? I
don't see why you need to kill everything and restart.

On Wed, Apr 17, 2019 at 2:51 PM hgn <[email protected]> wrote:

> I investigated a bit more.  You are right, obviously, they already are
> children.
>
> My scenario differs because I only look at the processes when stuff
> breaks and I needed to kill nsmd or non-session-manager.
>
> If you kill -9  this only nsm will quit and leave all child processes on
> their own. If you have hidden GUIs they will not be visible, even not as
> frozen xwindow.
>
> The problem is that I used the wrong kill command. There is a group
> kill, which works on the parent process (non-session-manager, the GUI in
> this case) and is invoked by giving the PID as negative number.
>
> kill -9 -5866
>
> That works.
>
>
> Am 17.04.2019 18:05 schrieb J. Liles:
> > NSM clients *are* children of the nsmd process (unless, as I think
> > you're implying, you've got things set up to use an external launcher.
> > But that only applies the first time the client is launched. When you
> > open the session again via NSM, the client will then be a child
> > process of nsmd).
> >
> > On Wed, Apr 17, 2019 at 8:22 AM hgn <[email protected]> wrote:
> >
> >> Hello list,
> >>
> >> when I start an application in NSM it is its own regular process,
> >> only
> >> under session management because the NSM env-var is present, and
> >> the
> >> rest happens via OSC.
> >>
> >> Sadly we don't live in a perfect software world so sometimes
> >> programs
> >> hang and freeze so bad that you need to close the session and
> >> reopen it.
> >> The problem is that this freezing prevents applications from
> >> getting
> >> closed correctly so it is not uncommon to have several processes
> >> running
> >> or hanging that you need to kill manually. Until everything works
> >> I
> >> want to have a pragmatic compromise.
> >>
> >> Is there a way to start every program from an NSM session as child
> >> process, so if NSM gets killed all programs get killed as well?
> >>
> >> If not is there a technical or logical reason against it?
> >>
> >> If no could that be implement trivially?
> >>
> >> yours,
> >> hgn
>
>
>

Reply via email to