rjk@mgweed.UUCP (Randy King) (03/22/85)
<><><> It seems that autobauding from 1200 down to 300 by means of a framing error is quite reliable when the 300-baud user types almost any character. Hence for years uucp info has contained strings like: login:--login: where the "--" implied "send CR". The reverse did not work, so the 1200-baud user would send a BREAK to upshift the remote to 1200; hence login:-BREAK-login: for the uucp info. Now I am finding that downshifting from 2400 baud to 1200 baud more closely resembles the 300->1200 upshift in that sending a BREAK is the more reliable way. It is not always the case, however. There are times when just a CR from my 1200 baud modem calling a tri-speed modem at 2400 baud will cause the port to fallback to 1200. Many times, however, a BREAK is required to cause the downshift. If you have had experience with 2400 baud autobauding, including upshifts and downshifts among the speeds from 110->2400, I'd like to hear about your findings. This would be important information for future releases of UUCP information from sites with >1200 baud modems. Randy King AT&T-CP@MG ihnp4!mogul!rjk
honey@down.FUN (code 101) (03/24/85)
CR is not the best character for generating a framing error. you want the receiver to see a 0 where it expects to see the stop bit. since CR is 00001101, the receiver sees the reversal of this bit string (rs232 is little-endian), with start and stop bits added, and every bit repeated, i.e., 001100111110000000011. note the long string of 1's where you would like to have a 0. the second byte will also fail to generate a framing error since it ends with the real stop bits. your best bet for generating a framing error is to use the ascii character that ends with the longest string of 0's: @, or 01000000. this is why you frequently see "" @ and in:-@-in: in uucp login scripts. @ also has the happy side effect of clearing out line noise, except in 4.2bsd where the convention was unilaterally changed. peter
wls@astrovax.UUCP (William L. Sebok) (03/25/85)
> Now I am finding that downshifting from 2400 baud to 1200 baud more closely > resembles the 300->1200 upshift in that sending a BREAK is the more reliable > way. It is not always the case, however. There are times when just a CR > from my 1200 baud modem calling a tri-speed modem at 2400 baud will cause > the port to fallback to 1200. Many times, however, a BREAK is required to > cause the downshift. I decided to take another tack here at astrovax. I installed code in my getty to add an option to explicitly autobaud on an initial carriage return. The code to do this on 4.2 BSD was posted on 23-Nov-83 by Jim McKie (jim@mcvax.UUCP). This has the welcome side effect that it prevents modems that echo dialing commands (or anything else) from getting into a conversation with getty, which interprets the echoing of the login banner as an attempt to login. -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls