Ethereal-dev: Re: [Ethereal-dev] Request: Change the allowed license of plugins

Note: This archive is from the project's previous web site, ethereal.com. This list is no longer active.

From: "Ronnie Sahlberg" <ronnie_sahlberg@xxxxxxxxxxxxxx>
Date: Thu, 21 Nov 2002 20:00:30 +1100
From: "Brad Hards"
Sent: Thursday, November 21, 2002 6:30 PM
Subject: Re: [Ethereal-dev] Request: Change the allowed license of plugins


> 1. This is a change to the existing license, and potentially needs to be
> approved by all copyright holders. What happens if someone says no?

That developers contribution must either be removed from ethereal or the
licence remains.
Cant see any other options.

> 2. Whether alternatives (such as the patent holders granting a restricted
> patent license for use of techniques potentially covered by patent) are
> compatible with the current license.

Software patents are for IMHO very good reasons not recognized in large
parts of the world.
(software patent != software copyright)

> 3. Whether the plug-in API is sufficiently well defined (in a
documentation
> sense, and also in a code stability sense) that this technique is
> practicable.

It probably is, with the addition of a dialog that tells everyone : "By
loading this plugin you are on your
own. Dont spam ethereal-dev with your whining if something bad happens."


More importantly, whatever is this going to acheive?
A protocol dissector plugin is not like a hw device driver.

All users have completely full control of exactly what input the plugin
gets, by text2pcap and shell scripts.
It is trivial to see what output a certain input resulted in by just looking
at what output the plugin generates.

Trust me I know. If you have a binary only protocol dissector, it is still
extremely trivial (but timeconsuming)
to recreate the ENTIRE protocol as known by the dissector.
Just try every single possible type of packet (using branch-and-bound
techniques plus some educated guesses) and compare it with what output the
dissector produces. Trivial.
Anyone that does not belive that could observe the empirical evidence in
that packet-srvsvc.c exists.


I will not oppose allowing for binary only plugins.
But the real question is,  If a binary plugin is released or whatever.


What would the plans be for the ethereal developers if a reverse engineered
source code dissector is
contributed to ethereal?
Should we then not accept it for inclusion based on the existence of a
closed source one?


I am not against it, I just dont see the point in distributing binary only
plugins since it is so trivial to
translate it back into a source code dissector again.