Session Management Proposal

[Ray Strode]

Hi Lars,

>  However, in any case, the only way to avoid potensial dataloss is to
>  save bothe the curent and the last saved version somehowe!
I think the XSMP actually specifies the use of temporary buffers and
things of that nature, but it was decided that this is too complicated
to get right in practice (see the urls in the footnotes of my original
message).

>  The instant logout could be dealt with width some uggly hack.... a
>  'debug' loginmode in the DM that makes every 'SaveYourself' message
>  result in an confimation dialog, possably showing the originating app
>  so users can identify broken apps. Such mode could have a timout so
>  it's not forgeten about and interupts an UPS shutdown on an unatained
>  machine whith an open souch session (possably disable itself on
>  locked screen to).
I guess that's an implementation detail the DM authors may want to add
to their program.  I think in general, though, if a program is doing
broken things, its not really the users responsibility to debug the
program (or even go through some complex process to identify it, if its
not immediately clear which program is causing the problem). 

>  Your last session was imediatly terminated by this program: xxxx Nowe
>  I have recived an imediate termination request agan from the same
>  program. is that OK? [Yes-terminate session] [Ignore request]
>  [Remove program from session]
This sounds like a nice feature, but it is an implementation detail,
so shouldn't really be standardized in the spec I guess.  It might be a
good idea to put it in there as an implementation recommendation.

--Ray