Slide & code update of http://heim.ifi.uio.no/michawe/romavisit/

[Michael Welzl < >]

Hi,

I made a tiny update to the slides (fixing the bugs we fixed when talking 
today), and adjusted the code to also do SACK.
NOTE: the tcp-socket-base.cc was also updated - these fixes were necessary to 
avoid bad interactions with existing SACK code.

Now, if you enable SACK, the sender parses the SACK blocks when receiving an 
ACK and removes the respective entries from its data structure.
The effect is … tiny, in the simulation scenario that I always use. I think 
it’s probably because it’s a DropTail queue, where you typically have a tail 
loss and SACK can’t help much - or I made a mistake somewhere  :-)

Next steps are:

- defining scripts to test this in depth and see if there are any bugs (if 
so, I’ll update but I won’t send an email about each and every update - you 
can look at the “last update” date at the bottom of the page… I guess a git 
repository would have been more efficient, but whatever… I’m not planning 
sooo many updates anyway!

- doing Giuseppe’s planned next steps that we talked about today

and that’s that. Else, the code (minus possible bugs) at least now does 
everything I wanted it to do !  :-)   I’m happy   :-)

Have fun,
Michael