13 Sep
2022
13 Sep
'22
10:36 p.m.
On Sep 13, 2022, at 3:18 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
DECnet/Linux definitely have issues. I would agree with what Bob writes.
But unless someone manages to trigger this through normal, interactive requests, the
mapper is also doing something slightly out of the ordinary. Which in fact seems likely,
and most of DECnet/Linux was tested by just running it against VMS as the debug tool. So
anything VMS isn't doing have not been properly tested at all.
Lots of reverse engineering behind the whole thing.
Sure, but then again NICE is fully documented (other than OS-specific requests) in the
architecture manuals; all that an implementer needs to do is implement what those specs
say. Unlike some other specs, DECnet in general was specified sufficiently carefully that
if you do what it says, it will work.
A good test would be to take an RSX or RSTS host and issue the NCP requests I described
the mapper issues, to see if that reproduces the issue. And if not, then it might be time
to trace the precise packets to see how they differ. As I said, there is nothing in any
way unusual about the NICE requests the mapper issues.
paul