hst@ifcom.UUCP (Hartmut Streppel) (08/14/89)
We have a PC/RT running AIX 2.2.1 connected to our LAN, consisting of a Sequent Balance, a Prime Exl 386, a Bull SPS 7, a Sun and sometimes other workstations, via TCP/IP. If we use rlogin to login from one of the machines mentioned above to the RT, and then try to use vi on the RT, vi fails giving the message: "Don't know enough your terminal to use visual" We have checked the stty settings, the environment variables, the terminfo files, the configuration of the pts; everything is ok. The funny thing is: when the system has just been rebooted, a single DEL before you call vi, might make it work. Is it possible that the driver gets some new information by this interrupt. Is this a vi problem? By the way, using telnet we have no problems using vi on the RT. Any help or ideas? Pleas email! Hartmut Streppel, InterFace Computer GmbH, Garmischer Str. 4, 8000 Muenchen 2, West Germany Tel: +49-89-51086-35, FAX: +49-89-51086-28, uucp: ...!mcvax!unido!ifcom!hst
jw@pan.UUCP (Jamie Watson) (08/16/89)
In article <396@ifcom.UUCP> hst@ifcom.UUCP (Hartmut Streppel) writes: >If we use rlogin to login from one of the machines mentioned >above to the RT, and then try to use vi on the RT, vi fails giving >the message: "Don't know enough your terminal to use visual" We have seen the same thing using rlogin from MicroVax and Vaxstation systems. We have also found that when this happens, logging out from the RT, and immediately rlogin again almost clears the condition, whatever it is. jw
sergio@squid.rtech.com (Sergio Aponte) (08/17/89)
I have had some luck by setting the pseudo-tty entries to be a vt100 type window. If you get one where term is set to dumb, vi does not work. To change these, use "devices" and go 1 by 1, or edit the /etc/ports file. There is an entry there for every port. After you change the file, there is a good chance you have to re-boot. (I personally use the "devices" command so that I don't get in trouble). It seems once the prot is "told" you have a dumb terminal, changing TERM or doing tset does not fix it. (!!) Of course, if anybody else knows better, by all means let me know what I have to do. Good luck. ------------------------------------------------------------------------------- | Internet: sergio@squid.rtech.com Sergio L. Aponte, MTS @ RTI | | UUCP: {sun,mtxinu,pyramid,pacbell,hoptoad,amdahl,cpsc6a}!rtech!squid!sergio | -------------------------------------------------------------------------------
brunner@bullhead.uucp (08/29/89)
>In article <396@ifcom.UUCP> hst@ifcom.UUCP (Hartmut Streppel) writes: >If we use rlogin to login from one of the machines mentioned >above to the RT, and then try to use vi on the RT, vi fails giving >the message: "Don't know enough your terminal to use visual" The error is due to an initial sanity check in the vok routine in ex_v.c, the portion of the ex/vi code which checks for terminal characteristics. If the global WCOLS hasn't been set prior to entry into this routine, then there is insufficient info to divide the screen buffer into lines. So, why hasn't WCOLS been set? Both rlogin and telnet make an effort to determin the TERM shell variable in the local environment and pass it to the remote host, where it may be reset during the remote host login. Check your .cshrc/.login on the remote host for settings, you may want to have your term variable host dependent, set in a switch statement in .login switch ($host) # ibmpa lan hosts case ibmsupt: set TERM = "ibmaed" breaksw case dumbbunny: set TERM = "dumb" breaksw case dialbunny: set TERM = "dialup" breaksw default: set TERM = "vt100" endsw P.S. Your mailage may vary with AIX, check ttys(5) for supported types and echo $term (or $TERM) on both ends of the pipe. An equivalent problem was reported as an APAR, xterm via telnet didn't force the remote host to use the local host's notion of the term shell variable. This is NOT a bug. Telnet/rlogin are general remote terminal routines, if the local setting for term were forced on the remote host it would be very inconvient, passing vt100 is always safe (even to TOPS and VMS hosts on the Internet), but the remote host may not support X or the Advanced Experimental Display or whatever. It is preferable to have a screen-oriented application (e.g., vi) bail out to line mode when the environment is either unsupported or not available during telnet/rlogin initiation. One can always set term in the login environment or "by hand". Eric Brunner, IBM AWD Palo Alto inet: brunner@monet.berkeley.edu or brunner%ibmsupt@uunet.uu.net uucp: uunet!ibmsupt!brunner (415) 855-4486
karish@forel.stanford.edu (Chuck Karish) (08/29/89)
In article <1934@ibmpa.UUCP> brunner@ibmsupt.UUCP () wrote: >>In article <396@ifcom.UUCP> hst@ifcom.UUCP (Hartmut Streppel) writes: >>If we use rlogin to login from one of the machines mentioned >>above to the RT, and then try to use vi on the RT, vi fails giving >>the message: "Don't know enough your terminal to use visual" >The error is due to an initial sanity check in the vok routine in ex_v.c, >the portion of the ex/vi code which checks for terminal characteristics. >If the global WCOLS hasn't been set prior to entry into this routine, >then there is insufficient info to divide the screen buffer into lines. >So, why hasn't WCOLS been set? Both rlogin and telnet make an effort to >determin the TERM shell variable in the local environment and pass it to >the remote host, where it may be reset during the remote host login. Check >your .cshrc/.login on the remote host for settings, you may want to have >your term variable host dependent, set in a switch statement in .login... Of course, the switch in .login is useless if you log in on a network port. I have my .kshrc prompt me for the terminal type. It doesn't work. The TERM variable is set to a name that has a good terminfo entry, but vi still won't come up in full-screen mode if I open the connection with rlogin. When I use telnet vi works fine. Chuck Karish karish@mindcraft.com (415) 493-9000 karish@forel.stanford.edu
vlruo02@dutrun.UUCP (Ge van Geldorp) (08/30/89)
In article <1934@ibmpa.UUCP> brunner@ibmsupt.UUCP () writes: >So, why hasn't WCOLS been set? Both rlogin and telnet make an effort to >determin the TERM shell variable in the local environment and pass it to >the remote host, where it may be reset during the remote host login. Check >your .cshrc/.login on the remote host for settings, you may want to have >your term variable host dependent, set in a switch statement in .login [deleted stuff] >P.S. Your mailage may vary with AIX, check ttys(5) for supported types >and echo $term (or $TERM) on both ends of the pipe. You seem to be talking about AOS, I can't comment on that since our machines runs AIX. However, I think the original poster was also running AIX, because the problem he describes sounds very familiar to me. As far as I can tell, the problem has nothing to do with the setting of the $TERM variable. It has to do with a bug in rlogind. If you use rlogin to login to an AIX machine, the window size (number of rows/cols) is not always transferred correctly to the AIX machine. You can confirm this by issuing the termdef -l and/or termdef -c commands. If the window size was not transferred correctly, both commands will return 0 (yes, that's zero, not an empty string). I suppose vi calls the termdef(3) (or the setupterm(3) routine, which in turn calls termdef(3)) routine on startup and gets this garbage back, after which it gives up. One of the systems from which I use rlogin to go to an AIX machine runs UTek, the Tektronix version of **ix. Recently, we upgraded the UTek version and we got a new version of rlogin. This version has a new flag, -w. The documentation says this flag is used to force correct transfer of the window size to some versions of rlogind based on the BSD 4.3 rlogind. The BSD 4.3 rlogind appearantly does not handle the necessary out-of-band data transfer needed for the window size exchange correctly. When I use this -w flag, vi on the RT works correctly... (BTW, if the problem was really with the $TERM variable, vi wouldn't complain "Don't know enough about your terminal to use visual" but would say "xxxx: Unknown terminal type". Try it.) I reported my problems with rlogin/vi several months ago to IBM but haven't received a fix yet. Now, would somebody at IBM please take a good look at it please please please? Ge van Geldorp. (ge@dutlru2.tudelft.nl)