>Once upon a time Carl Oppedahl shaped the electrons to say...
>>This init string is sent to the modem every time the router is reset, and
>>every time the port is reset, and every time the modem loses carrier. The
>
>wrong wrong wrong
Well, as a spokesperson for Livingston you are surely giving a nice warm
welcome to a new owner of a Livingston router. Maybe next time I'll buy a
Cisco or an Ascend ... in fact I am still within the return period, maybe I
will return this Livingston and tell them I would like a different brand of
router ...
>It is sent on boot and on MANUAL reset of the port. NOT after each call.
I see. Well, since you work for Livingston and are privy to such things,
can you tell me where, in any of the router documentation, it lists the
circumstances in which the router re-initializes the modem? More to the
point, can you point to the place where an *exhaustive* list is provided?
I'd like to look there and see that it does not do it per-call.
Keep in mind that my posting was about dialout, not dialin. Are you sure
the router doesn't re-init the modem upon dialout?
>>bad thing about this string is the &W, which rewrites the flash EPROM in the
>>modem. The problem is that there is a limited life for any flash EPROM, a
>>limited number of times it can be rewritten. This init string wastefully
>>consumes some of that limited life every time the init string is executed.
>
>This argument is moot as the data it was based on is incorrect.
No, not moot at all. Even if the router's (undocumented) behavior is that
it doesn't re-initialize the modem for each dialout call, that still leaves
a situation where the router uses up some of the modem's limited EEPROM life
every time there is a manual port reset and every time the router is
rebooted. And for no good reason, as far as I can see.
>>It would also be desirable if the init string contained a command to fix the
>>DTE rate. In my setup I am using the modem for dialout, and I want to be
>
>The correct string for a USR, from the manaul, is AT&F1s0=1&W
No, that's not a correct string for a USR, not for any process that is
repeated from time to time. &W should be done only on an as-needed bases,
not as a routine step. The EEPROM has a limited life.
>>sure I get error correction (&m5) and fixed DTE baud rate (&b1). So here is
>>a different init string:
>
>&F1 should provide both.
I am sorry but you are mistaken about this. The &F1 default does not
provide an assurance of error correction, because it only defaults to &M4.
And the modem's behavior with &M4 is that it will quite readily establish a
non-error-corrected session on the slightest level of line noise.
The modem string you are so strongly defending, which uses &F1 but not &M5,
is suboptimal for dialout links to other places. Here's why. Consider the
consequences if line noise happens to make a V.32/MNP4 error correction
impossible. Then the two modems will handshake to a non-error-corrected
session. All manner of errors can creep in, depending on the nature of the
serial session. If it's PPP, the result will be higher-level packet resends
and resultant degraded throughput. If it's not PPP, e.g. if it is some sort
of ascii terminal session, then the entire session will be adversely
affected by line noise.
&M5 is much, much better, in the case where it is known that the modem at
the other end supports error correction. With &M5, when the router dials
out, if error correction cannot be negotiated, the modem will report "no
carrier" and a redial will be done, until an error-corrected session can be
established between modems. In this way the router causes the modem to
redial until a clean line can be obtained.
>>If there are other modem init strings in PortMaster Version 3.4.2L that
>>perform &W (I haven't checked) I suggest they be revised as well.
>
>All of them do. And it is deliberate. We're not going to change them.
Well, I am very sorry that your stubbornness will lead to the EPROMS in your
user's modems being used up sooner.
Let's suppose you were correct when you thought that &F1 provided &M5 in
that modem. Then instead of using AT&F1&W\r, why not simply use AT&F1\r?
This would achieve all of your goals, and would not use up the EEPROM's
life. What do you lose by not doing the write to the EEPROM?
--- Carl Oppedahl, Oppedahl & Larson, patent law firm http://www.patents.com/ has hundreds of pages of answers to frequently asked questions on patent, copyright, and trademark law