Roland Mainz wrote: > Bill Shannon wrote: >> Glenn Fowler wrote: >>> On Wed, 12 Mar 2008 17:33:11 -0700 Bill Shannon wrote: >>>> Bill Shannon wrote: >>>>> Glenn Fowler wrote: >>>>>> On Mon, 10 Mar 2008 20:03:30 -0700 Bill Shannon wrote: >>>>>>> Who's invoking "tty -l" when I use ksh93? It's not in any of my >>>>>>> dot files. >>>> According to truss, this message is coming from ksh93 itself. >>> if its the builtin tty from -lcmd it would show up in the same process >>> there must be a tty -l call in some script somewhere >> I haven't found it. How can I find it if it's using a builtin? > > All AST tools (AST="Advanched Software Technology", which is more or > less a superset of the upstream package which contains ksh93) support > the "--man" option, e.g. you can test for the builtin version of "tty" > via $ tty --man # ; the official ksh93 way to test for it is $ builtin | > fgrep tty # to check whether the "tty" builtin is enabled by default and > whether it's bound to a path...
That's not what I meant. What I meant was how could I find out that it was an invocation of the builtin tty command that was causing the message? >> It's not in the new /etc/ksh.kshrc. > > What about /etc/profile ? Nothing there. >> The only occurrences of the word "tty" in my dots files are: >> >> $ grep -w tty .profile .kshrc >> .profile:else # normal tty login >> .kshrc:if [ `tty` != /dev/console ] > > AFAIK I found the problem; the builtin "tty" has a small glitch which > causes it to always print "not on an active synchronous line": > -- snip -- > $ tty > /dev/pts/2 > not on an active synchronous line > -- snip -- Uh, ya, that would do it! > [snip] >> I know everyone's fascinated by this strange problem, but I'm sure it >> will be easily found when someone fires up a debugger on ksh93. But >> really, it's not the problem I'm worried about. The problem I'm >> worried about is the problem I filed the bug report for! Anyone looking >> into that? > > I tried to look into that problem... the trouble is that I can't > reproduce it on my side... ;-( Based on my previous experience with ksh, I suspect it's a race condition. Which means it might be very environment dependent. If you can suggest some truss or dtrace or whatever commands that might tell us what's going on, I'd be happy to run them.
