ANNOUNCEMENT: Live Wireshark University & Allegro Packets online APAC Wireshark Training Session
April 17th, 2024 | 14:30-16:00 SGT (UTC+8) | Online

Wireshark-dev: Re: [Wireshark-dev] Further HTTP woes...

From: "ronnie sahlberg" <ronniesahlberg@xxxxxxxxx>
Date: Fri, 8 Sep 2006 09:43:43 +1000
That would solve the problem for YOU   but would be the wrong way to solve the issue.

We dont want dissectors to start implementing what is essentially tcp reassembly and tcp session tracking in the dissectors themself since this belongs in the tcp dissector   not in the application dissector.


While this issue is most commonly occuring with HTTP based protocols   there are other protocols as well that
transfer only one single PDU per tcp session and which sometimes do not provide a pdu length in the header

Thats another reason why the solution is to add the missing semantics to TCP and not reinvent a solution private to the http layer.
Implement the enhancement once and several protocols will benefit.




On 9/8/06, Bryant Eastham <beastham@xxxxxxxxxxx> wrote:
And what of the solution to pass a zero-data-length buffer to the subdissectors and use a conversation?