the sorry saga of the talloc soname 'fix'

[Sam Liddicott]

* Stefan (metze) Metzmacher wrote, On 09/07/09 11:21:
> None of this 4 patches passes make test in the a standalone build,
> so there's nothing that can be judged :-(
One of the tests was supposed to fail:
    /* The ugly talloc de-child-looping code will delete p2's reference
       leaving p2 having a dangling pointer. p2's reference should remain */

It still requires some better memory loop management code to be written,
the test was failing in anticipation of that.
I'll fix up that test to expect current behaviour until we have the loop
discussion some other week after we all had a rest.

I'm re-factoring the patches as you asked and will post them soon.

Sam