gandrews@netcom.COM (Greg Andrews) (02/18/91)
In article <1991Feb17.193955.25078@rwwa.COM> witr@rwwa.COM (Robert W. Withrow) writes: > >>Protocol support for uucp (or kermit or x/ymodem) is available only >>in two modes: PEP and V.32/MNP. "Direct V.32" does not have protocol >>support. > >Everyone I talk to tells me something different on this (including >Telebit tech support! Everyone agrees that spoofing is performed >in PEP mode, but no-one can agree on which other modes do spoofing. > Well, the above is the complete and whole truth, coming from someone who works for Telebit Tech Support. If you think about it, spoofing sends an acknowlegement back to the sending computer from the modem instead of waiting for it to come back from the receiving computer. If there is no assurance that the data will reach the receiving modem intact, much less the receiving computer, then the modem shouldn't do spoofing. Therefore, in "V.32 Direct" mode (without error correction) the Telebit modems don't do spoofing. PEP mode is obvious. V.32/MNP is the other mode, for the above reasons and the fact that Microcom were ready to license manufacturer-specific extensions to MNP. The CCITT doesn't seem to be prepared to do that, so there's no spoofing in LAP-M mode at present. > >>OTOH, the modem-to-DTE flow control can still have its little say. > >Not when you have disabled *all* flow control (as I stated in my >post). Depending on the modem (and I *dont* use Telebit modems), >there seems to be enough buffer space to support UUCP (with *all* >DTE-Modem flow control turned off) without lost packets, even when the >DTE-Modem data rate is greater than the Modem-Modem data rate. This >is a purely empirical estimate on my part thought... > Actually, you mentioned "DTE-to-modem" flow control. I didn't know you meant modem-to-DTE as well, so I addressed that issue... Well, if your modem has the RS232 speed unlocked (S66=0) then you'll probably be okay on that most of the time. If you run with a locked interface, then your dialup interactive sessions can have trouble with slow speed connections (2400 and under). As for the buffering, yes the modems should have enough to buffer uucp transfers that use a window of 3 64-byte packets. I've seen messages about how larger packet sizes are possible under uucp. If the packets are big enough, three of them could overflow the modem's buffer. -- .-------------------------------------------. | Greg Andrews | gandrews@netcom.COM | `-------------------------------------------'