DB> I *assume* (since V.34 isn't clear at *all* on this point), that the
DB> *call* modem sends MP with *all* bits set according to what it supports,
DB> and the *answer* modem returns MP with only *one* bit set, according to
DB> which one it wants to use (of those supported by the call modem).

DB> The whole MP/MP' sequence exchange is *very* ambiguous and unclear,

Yeah, classic rather fogged up specification.

DB> and you'd probably need the rationale, working notes etc.
DB> of the ITU-T working group in order to know the real answer.
DB> Unless it's covered in a previous ITU-T spec.

Craig Ford in any of the international echoes like USR_Modems or
Comms could likely fill some of that detail in. Tho his style is
not to usually quote directly from that sort of stuff, and he has
made some comments at times that he isnt allowed to.

DB> Look at table 20 (page 34), the description of
DB> the MP sequence Type 0 - see for bits 29 and 30, it says:

DB> Trellis encoder select bits:

DB> 0 = 16 state, 1 = 32 state, 2 = 64 state, 3 = reserved for ITU.
DB> Receiver requires remove-end transmitter to use selected trellis
DB> encoder.

DB> Who here is the "receiver"?  The receiver of the MP sequence, the
DB> demodulation (receiving) side of the modem sending the MP sequence,

Yeah, thats the most obvious question, and likely they
are actually SEPARATELY negotiated in each direction,
so you can and do get 64S one way and 16S the other.

DB> If that last line is to be taken literally, and the
"receiver" is the
DB> demodulation side of the modem sending the MP sequence, then it would
DB> seem to indicate that all three forms of trellis coding are *mandatory*.

DB> How many wires are there in a telephone cord - two or four?

Two, but thats not relevant coz a full duplex modem like V34
normally is used just uses a hybrid to do the 2wire to 4 wire
conversion at each end and the modem can treat each separately.
And clearly does if you can get a 64S/16S session.

DB> How difficult would it be to make a split cable that ran between
DB> two modems to enable them to negotiate a connection directly, but
DB> have the split section run off to one or more A/D converters and
DB> thence into a PC that could capture the conversation between the modems,

Thats 4wire mode which some V34 modems support and some dont.
It shouldnt make any difference to that negotiation detail tho.

DB> allowing us to see once and for all exactly what the two modems were saying

You cant monitor the traffic over the link at the level with what we have
anyway. Even the fine detail step by step spell out of the negotiation
that the M34F provides if you want it doesnt go down to that fine a level.
Or atleast what undocumented commands we know about dont.
@EOT:

---