The note describes the validation tests
in test-suite-sack.tcl.

SACK1 and SACK1Z:
The test "sack1" shows Fast Recovery with SACK when a single packet
is dropped.  Note the burst of packets when the ACK for packet 26
is received at time 2.6.  Test "sack1z" shows the same test with
maxburst_ set to 4 packets.

SACK1A and SACK1AA:
Test "sack1a" shows Fast Recovery with SACK when three packets are
dropped from a window of data.  Notice the burst of packets at time
1.65.  Test ""sack1aa" shows the same test with maxburst_ set to
4 packets.

SACK1B and SACK1C:
Test "sack1b" shows Fast Recovery with SACK when many packets are
dropped from a window of data.  Test "sack1c" is a variant of test
"sack1b" in that is shows correct behavior when the last packet
sent before a Fast Retransmit is dropped.

SACK3:
Test "sack3" shows, for the bottom connection, a Fast Recovery with
SACK when packets are dropped from a connection with a small
congestion window.

SACK5 and SACK5A: 
Test "sack5" shows the burst of packets from the bottom connection
sent at time 3.5.  Explain!!  For the top connection, this test
shows the behavior when half the packets are dropped from a large
window of data.  Test ""sack5a" shows the same test with maxburst_
set to 4 packets.

SACKB2:
Both connections in this test show show a Fast Retransmit, followed
by a retransmit timeout for a retransmitted packet that is itself
dropped.  Because the bottom connection's congestion window is
relatively large when the initial Fast Retransmit occurs, it begins
the retransmit timeout with a reasonably large value for ssthresh,
and recovers fairly quickly.  Because the top connection's congestion
window is small when the initial Fast Retransmit occurs, it begins
the retransmit timeout with a smaller value for ssthresh, and
recovers more slowly.

SACKB4 and SACKB4A:
Test "sackB4" shows the burstiness when SACK TCP recovers from
two packets dropped early in a large window of data.
Test "sackB4a" shows the same test when "maxburst_" is set to
four packets.

FALSEPIPE:
Test "FalsePipe" shows that, during fast recovery, the pipe_ value
used by SACK can be over-estimated if partial ACKs are trigered by 
packet reordering instead of by packet loss.

DELAYEDSACK:
The "delayedSack" test from the v1 test suite should be included.

TIMERSSACK:
The test "timersSack" needs to be checked.

PHASE:
The phase tests were deleted as unnecessary to the validation test.
