too many layers in selftest

[Douglas Bagnall]

On 16/02/20 2:31 pm, Jeremy Allison wrote:

> Hmmm. Does you end up seeing dog faces everywhere in the
> code ? That's a strong argument for us living in a simulation :-).

Well, now you mention it, I am trying to grep for dog faces, and
finding the tool woefully unsuited. The best I have found is the
owl-like:

lib/ldb/pyldb.c:        if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|O",

and the last figure here:

python/samba/kcc/graph_utils.py-                        o
python/samba/kcc/graph_utils.py:    pass: o-o=o  o=o   (|)             fail:  o-o
python/samba/kcc/graph_utils.py-            `o          o                     `o'

So maybe we are safe.

On the other hand, I am not sure your argument works for samba tests.
When you're already in a fake environment, in cwrap, in a container,
on virtual machines, what difference does one more layer of simulation
make?

Douglas