On Tue, 2008-02-05 at 06:51 -0800, John Plocher wrote:
> Laszlo (Laca) Peter wrote:
> 
> > gnome-session waits for all postrun scripts in the JDS_wait
> 
> 
> A few questions/comments ...
> 
> jmp-1) Brand and organization names like JDS should never be used in
> architectural level artifacts - rather a descriptive word should
> be used instead.

Understood.  JDS here refers to the desktop environment, in contrast
with CDE or KDE.

> jmp-2) There should be a conceptual distinction between graphics system
> startup, login screen, and desktop startup.  This would allow the user
> or admin to use KDE (or CDE or ...) instead of GNOME as their desktop.

Right.  That's why it's gnome-session that waits for postrun
and not X or gdm.  Note that this doesn't normally happen: by
the time the user logs in the postrun jobs are finished on most
systems, only some lower end machine will ever have to wait.

> If the postrun stuff is tied to only the gnome desktop, yet it does
> things for the other two conceptual areas, we have problems.  

postrun itself is not tied to the GNOME desktop, but the jobs
submitted to postrun are.

> In the same
> way, if postrun is a JDS-only thing, what mechanism would KDE (...) use?

KDE could (in theory) use postrun and define its own class of jobs
that it may need to wait for.

> jmp-3) How does this relate to the work being done in IPS, which is also
> trying to fix/eliminate postinstall scripts?

Stephen Hahn has already responded to this.

Laca



Reply via email to