["Peter J. Creath" <peter-ethereal@xxxxxxxxxx>]

> we could add support for MAC-address-based dissector tables, to 
> handle protocols that, like ISL, involve non-802.3 packets sent to 
> particular MAC addresses

You're right:  a MAC table is probably better (faster) for things 
like ISL, although ISL has a fair number of MAC addresses allocated 
to it -- a table might be a bit cumbersome.

The packets I'm examining use a large range of private MAC addresses. 
The range is large enough that a simple MAC table wouldn't work very 
well -- it's far easier to match programmatically.

And there's no guarantee _all_ the traffic will be non-standard, so 
the DOCSIS-style switch isn't so useful here.  Inserting a heuristic 
test allows someone to interleave standard and non-standard 802.3 
frames, assuming the heuristic dissectors are sufficiently 
conservative.

Should I formally re-submit my patch to the list (I realized I didn't 
use the usual "[Patch]" marker in the subject line), or do you not 
like the approach?

	-P