Ethereal-dev: Re: [Ethereal-dev] 0.10.10 next week?

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 This list is no longer active.

From: Lars Roland <lars.roland@xxxxxxx>
Date: Sat, 05 Mar 2005 06:37:46 +0100
There's is one good reason for a dissector or what else, to be a plugin.
If it is experimental or still on an early development stage, you can have an option in the installer.

I agree, we have a lot of plugins stable enough to be part of ethereal itself.
We should depluginize them.
So we come to another discussion I want to start here.

How about grouping dissectors together in sub directories of epan/dissectors?
I don't like to have 700 files in just a single directory.
I'd like to suggest following groups for dissectors:
- Layer 2 Protocols
- Layer 3 Protocols
- Layer 4 Protocols
- ASN.1 based protocols
- text based protocols, e.g. HTTP, SIP, MGCP (depluginized), ...
- dcerpc dissectors
- aim dissectors
...



ronnie sahlberg schrieb:
do we need plugin support?   it is a serious question.

the plugin api changes rapidly enough to cause pain and very frequent
recompilations for people maintaining out-of-tree dissectors/plugins
for several for-the-public-uninteresting or weirdo protocols.
i use several of those myself.

so   in light of that,  that they have to patch and recompile (or i
would assume would be the normal case :   resync the plugin to current
ethereal once a year or so)   does having plugin suipport actually
matter?
does plugin really reduce the out-of-tree maintenance cost anything at
all compared to maintaining an out-of-tree normal (not plugin)
dissector?
i think not.

ergo   support for plugin dissectors are semi-useless.


for several such weirdo out-of-tree protocols i sometimes encounter,
the maintainers have given up even trying to use plugins since there
is no saving in cost of maintenance using plugins.   instead   once a
year or so  the dissector is forward ported to the current version of
ethereal. and that special new version of ethereal is used instead for
a year or so.


so, in my experience,  plugin support does not provide any practical
benefit at all for maintainers of out-of-tree protocols which leads to
the quite serious question :  what is the benefit of plugin support?

end of rant.




On Fri, 04 Mar 2005 20:22:15 -0800, Guy Harris <gharris@xxxxxxxxx> wrote:

ronnie sahlberg wrote:

why not call it 1.0

Because I don't want somebody to complain when I change the way strings
are handled so that we can cope with strings in messages being in
various character sets?

Calling it 1.0 is fine, I guess, as long aa the internal API for
dissectors, etc. can still change without people having any chance to
complain that their old plugins no longer work.  (Yes, I know, some
people are *already* hoping to build plugins that work with multiple
releases.)



_______________________________________________
Ethereal-dev mailing list
Ethereal-dev@xxxxxxxxxxxx
http://www.ethereal.com/mailman/listinfo/ethereal-dev