On Tue, Dec 18, 2012 at 01:43:38PM +0000, Frank Shearar wrote:
> On 18 December 2012 01:58, David T. Lewis <lewis at mail.msen.com> wrote:
> > On Mon, Dec 17, 2012 at 01:15:12PM -0500, David T. Lewis wrote:
> >> On Mon, Dec 17, 2012 at 02:59:23PM +0000, Frank Shearar wrote:
> >> > On 17 December 2012 14:12, David T. Lewis <lewis at mail.msen.com> wrote:
> >> > >
> >> > > Setting useOldNetwork to true will cause IPv6 support to be disabled. Too
> >> > > many double negatives I suppose ;)
> >> >
> >> > That makes sense, but I'm still confused, because both my little
> >> > release script and ReleaseBuilder-fbs.85 set useOldNetwork to true.
> >> >
> >> > But NetNameResolver class >> #initiaize has "Smalltalk
> >> > addToStartUpList: self", so could actually set useOldNetwork BACK to
> >> > false even though the image was saved with value true?
> >>
> >> Yes that might be what is happening. And/or it may be that the preference
> >> needs to be initialized. I am away now but will try to have a look at it
> >> again later today.
> >
> > Frank, I really have to apologize - I gave you a bunch of wrong information.
> >
> > You were correct all along to do "NetNameResolver enableIPv6: false". This
> > sets the current value of the pragma preference for "Enable IPv6 and new
> > network support". So setting this to false in ReleaseBuilder is the right
> > thing to do.
> >
> > Contrary to what I said earlier, setting "NetNameResolver useOldNetwork: true"
> > does *not* disable IPv6 permanently. It is a temporary flag that will be
> > reset at image startup time based on the setting of the enableIPv6 preference,
> > and also based on whether the VM provides the new IPv6 primitives, which
> > is checked at image startup time.
> >
> > I do not see anywhere that NetNameResolver class>>initialize would be
> > sent in normal use, so I do not think that is the problem.
> >
> > Bottom line: If you can do "NetNameResolver enableIPv6: false" prior to
> > saving the image, this *should* set the preference correctly and it *should*
> > cause the image to be restarted in the correct state regardless of VM.
> >
> > I really apologize for the (my) confusion. Next time you publish the
> > release candidate, please do the "NetNameResolver enableIPv6: false"
> > exactly as you had originally intended, and let's see if it comes out
> > right this time around.
> 
> Thanks for everything, Dave. I've released a new version at
> http://ftp.squeak.org/4.4/Squeak4.4-RC3.tgz so let's see how that one
> works.

The preferences are set correctly now. But I took a quick look this morning
and I think I saw a case where the useOldNetwork flag got set back to false,
so I am afraid that there must be a bug in the initialization somewhere just
as you and St?phane had suggested earlier. I can't see where the problem is
happening, but something is not right. I'll try to take a better look at it
later today or tomorrow.

:-/

Dave