Re: orderlyShutdown question

Hi Thomas, I see throughout the Runner code some support for registering and
notifying deadManSwitches when a vat dies (becomes redirected to a
DeadRunner). However, no one ever calls Runner.addDeadManSwitch(), so no
deadManSwitches are ever getting registered. It looks like the intention is
that HeadlessRunner keep all outgoing refs in its myDeadManSwitches
collection, which, as the comment says, would then need to be a weak
collection. Without doing so, not only do we get the problem you observed,
we also fail to send __reactToLostClient messages on broken outgoing links.

IIRC the redirector logic, making this change will also repair the problem
you observe without further effort. But I'm not sure.


On Fri, Jul 2, 2010 at 8:28 AM, Thomas Leonard <
tal-v5nx5w6akNyLE8xUarVfuPLx9OUvmyODWmv/[email protected]> wrote:

> On Fri, 2010-07-02 at 08:14 -0700, Kevin Reid wrote:
> > On Jul 2, 2010, at 2:37, Thomas Leonard wrote:
> >
> > > I'm having a bit of trouble with shutting down vats. Here's some test
> > > code:
> > >
> > > def makeVat := <unsafe:org.erights.e.elib.vat.makeVat>
> > > def seedVat := <elang:interp.seedVatAuthor>(<unsafe>)
> > >
> > > def test() {
> > >     def [_, vat] := seedVat("5")
> > >     println(`created vat $vat`)
> > >     println(`shutting down vat $vat`)
> > >     def result := vat<-orderlyShutdown("reason")
> > >
> > >     return when (result) -> {
> > >             println(`shutdown complete: $result`)
> > >             test()
> > >     }
> > > }
> > >
> > > test()
> > >
> > > interp.blockAtTop()
> > >
> > >
> > > The problem seems to be that once the vat is shut down, it doesn't
> > > respond to __whenMoreResolved, so the caller is never notified.
> >
> > Is that a near or far reference to the vat object?
> >
> > It's OK to use vat objects from other threads, so try having a near
> > reference instead. (And if that helps, then we can discuss the right
> > way to have that reference at hand.)
> >
> > That said, it seems to me that having a reference into a dead vat (in
> > general, whether the ref is to a vat object or just some other object
> > in that vat) ought to resolve/WMR as a broken reference.
>
> It's near, but using "." instead of "<-" doesn't make any difference
> (the result is still a promise). Here's a slightly simpler test-case:
>
> def makeVat := <unsafe:org.erights.e.elib.vat.makeVat>
> def seedVat := <elang:interp.seedVatAuthor>(<unsafe>)
>
> def [_, vat] := seedVat("5")
> println(`created vat $vat`)
> println(`shutting down vat $vat`)
> def result := vat.orderlyShutdown("reason")
> println(`shutdown result: $result`)
>
> when (result) -> {
>        println(`shutdown complete: $result`)
> }
>
> interp.blockAtTop()
>
>
> $ rune shutdown.e
> created vat <Vat newSeedVat in <runs in newSeedVat>>
> shutting down vat <Vat newSeedVat in <runs in newSeedVat>>
> shutdown result: <Promise>
> [hangs]
>
>
>
> --
> Dr Thomas Leonard
> IT Innovation Centre
> 2 Venture Road
> Southampton
> Hampshire SO16 7NP
>
> Tel: +44 0 23 8076 0834
> Fax: +44 0 23 8076 0833
> mailto:tal-v5nx5w6akNyLE8xUarVfuPLx9OUvmyODWmv/[email protected]
> http://www.it-innovation.soton.ac.uk
>
> _______________________________________________
> e-lang mailing list
> [email protected]
> http://www.eros-os.org/mailman/listinfo/e-lang
>



-- 
Text by me above is hereby placed in the public domain

   Cheers,
   --MarkM

_______________________________________________
e-lang mailing list
[email protected]
http://www.eros-os.org/mailman/listinfo/e-lang