Looks, that my message to the plugin-discussion got lost in the last
week.
Here it is back again:
Guy Harris wrote:
> > The "other dissector" is another PLC-protocol ontop of COTP.>
> Does it run atop COTP atop full CLNP, or just atop the inactive subset?
It's using the inactive subset. What a luck, because I have to implement
a protocol driver for WinNT. So the layer 3 reduces to a single
0x00-byte ;-)
[Changing all plugins]
> That's one way of doing it - but another way might be to allow plugins
> not to have "protocol", "filter_string" or "dissector" values, with
> plugins lacking those values calling "dissector_add()" or
> "heur_dissector_add()" in their "plugin_init()" routine, registering
> their dissector, or dissectors, as appropriate.
That's another way of doing it. But we would loose the option to
enable/disable plugins if the plugins use this "sidedoor".
Personally I would prefer the way with the return value even it means
to touch all existing plugins. Hey, we are still in the 0.x releases and
we broke the plugin-API already with 0.8.8:
The introduction of "heur_dissector_add()"
changed the interface defined in plugins/plugin_table.h. So *all*
plugins under Win32 must be recompiled for this new release.
Gerrit
--
Dr.-Ing. Gerrit Gehnen Abt. LT
ATR Industrie-Elektronik Textilstra�e 2, 41751 Viersen, Germany
Tel. (+49)2162 485-362 Fax (+49)2162 485-100
http://www.atrie.de