stjohn@penelope.Oswego.EDU (Dave St. John) (10/09/90)
I am trying to get tcsh to run under AIX version 3. I am compiling with the BSD compatible libraries and defining things BSDish but when I run the shell input is all messed up. I type in the command ls and have to press return several times to get the command to appear and execute. Also commands such as more don't behave properly. When such a command is used it is sent signal 22 (TTOU) and is stopped. This occurs with several other output commands but some work occasionaly such as cat. I tried compiling with System V flags set and a combination of the BSD and System V flags but nothing produces a good result. Under System V the output from commands such as ls is all messed up with linefeeds not interpreted properly. Basically everything is messed up. I would appreciate any information about AIX specifics that need to be taken care of or if anyone has done this already some pointers in the right direction. Thanks in advance Dave St. John stjohn@oswego.oswego.edu
steve@qe2.awdpa.ibm.com (Steve DeJarnett) (10/09/90)
In article <1990Oct8.185748.15133@oswego.Oswego.EDU> stjohn@penelope.Oswego.EDU (Dave St. John) writes: >I am trying to get tcsh to run under AIX version 3. I am compiling with the >BSD compatible libraries and defining things BSDish but when I run the shell >input is all messed up. I type in the command ls and have to press return >several times to get the command to appear and execute. Probably something like 4 times per input line, right?? Actually, I think it's any 4 characters, but return is always the most common candidate. >Also commands such as more don't behave properly. When such a command is used >it is sent signal 22 (TTOU) and is stopped. This occurs with several other >output commands but some work occasionaly such as cat. Not quite sure why this occurs, but, as I recall, we're seeing something very similar here (I suspect it's the same thing, actually). >I would appreciate any information about AIX specifics that need to be taken >care of or if anyone has done this already some pointers in the right direction. There's a new copy on tesla.ee.cornell.edu that you might want to look at. There's a binary for the RISC System/6000 (RS/6000) there that you may or may not want to pick up (depending on how paranoid you like to be about picking up binaries from anonymous ftp sites). There's also source with the relevant AIXv3 defines in it. The comment relevant to the above-mentioned problem is: /* * If someone could explain this to me... * On the RS6000, CBREAK behaves like ~ICANON with VMIN=4 and VTIME=0! * So for this module, we use the SYSV tty driver stuff */ The version there is 5.19, which, apparently, is not a Paul Placeway version. It's an incremental release (non-official) done by some people at Cornell. It should solve most of your problems, though. >stjohn@oswego.oswego.edu Hope that helped. Steve Steve DeJarnett Internet: steve@ibminet.awdpa.ibm.com IBM AWD Palo Alto UUCP: uunet!ibminet.awdpa.ibm.com!steve (415) 855-3510 IBM VNET: dejarnet at ausvmq These opinions are my own. I doubt IBM wants them.......
wlm@arnor.uucp (10/09/90)
There is a version of tcsh on gatekeeper.dec.com which works correctly on machines which have posix-ish features. It is very easy to make this work on an RS6K; the only gotcha is that chars must be chars and not ints - this is re a feature of this version of tcsh (5.19). Bill Moran
gwyn@smoke.BRL.MIL (Doug Gwyn) (10/11/90)
In article <1990Oct9.052828.11184@panews> steve@ibmsupt.UUCP (Steve DeJarnett) writes: > Probably something like 4 times per input line, right?? Actually, I >think it's any 4 characters, but return is always the most common candidate. 4 is a significant "magic number"; it's the value of the default EOT character. System V multiplexes the EOT character in the termio structure with the non-canonicalizing VMIN field. If whoever ported the software failed to take this into account, they probably left the EOF character in the termio field when switching to non-canonicalizing mode. The bug should be easily locatable in the source code..