["Olivier Biot" <ethereal@xxxxxxxxxx>]

This trace is incomplete (content-length and reassembled length) don't
match so the TCP reassembly code never performs subdissection on the
HTTP reply.

A fix should have the means of reassembling (and dissecting) HTTP
headers and HTTP entity *separately* (both in requests as in replies).
I'm not sure whether the TCP reassembly code would allow us to do this
today.

Regards,

Olivier
----- Original Message ----- 
From: "Joerg Mayer"

As mentioned already on this list:
There seems to be a bug in current (t)ethereal where *sometimes*, http
responses are flagged as tcp desegmentation packets. I've attached a
capture that shows this behaviour. The source is current cvs and all
http and tcp bells and whistles are turned on.