[Standards-JIG] Re: The Ack Hack.

Matthias Wimmer m at tthias.eu
Sat May 13 21:44:37 UTC 2006


Hi Michal!

Michal vorner Vaner schrieb:
> No, it just forces TCP to be sending something and discover it is dead.
> Connection that does not send anything will look like alive even if long
> time dead.
>   
That's not always true. TCP knows (optional) ways to discover broken 
connections even if nothing is transfered.
>
>>
>> Some people told me that there are difficulties working with tcp under 
>> different OSes. That you sometimes can't get the info about what was sent and 
>> what was not sent.
>>
>>     
> No, this is the same on all OSes. You put data to the socket and expect
> it to get there. And TCP is designed to either deliver it or die. So, if
> it ends OK, you are sure everything got there. If it dies in unclean
> way, you shuold consider the whole stream as invalid. TCP was not
> designed to keep long-term connection, but to transfer data and
> terminate.
>   

The problem with have "with TCP" is not really a problem with TCP but 
with the design of a "socket" which is our typical interface to a TCP 
connection. TCP knows exactly what has been transfered (and ACKed) and 
what hasn't. It's just the socket interface, that does not give this 
information to the upper layers.
Therefore you can say that this is a problem with the OSes (which 
implement the socket) and not with the TCP protocol.


Matthias


-- 
Matthias Wimmer      Fon +49-700 77 00 77 70
Züricher Str. 243    Fax +49-89 95 89 91 56
81476 München        http://matthias.wimmer.name/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.jabber.org/pipermail/standards/attachments/20060513/b5f6746a/attachment.html>


More information about the Standards mailing list