Re: Problem with groff's configure script on z/OS Unix

[Akim Demaille]

| > Got the latest from CVS and got it over to z/OS UNIX.  Unfortunately
| > it does not build.  First, configure died with the error:
| >
| > ./configure 772: FSUM7341 bad file descriptor "5"
| >
| > That is when I run it from a script - but then when I run it from
| > the command line, I don't get this error.  This is probably another
| > OS/390ism that should be looked into (but I haven't seen it before
| > with other configure scripts).
| 
| This means it dies on the following snipped from groff's configure
| script built with autoconf 2.52:
| 
| ======================================================================
| 
| {
| cat <<_ASUNAME
| ## ---------- ##
| ## Platform.  ##
| ## ---------- ##
| 
| hostname = `(hostname || uname -n) 2>/dev/null | sed 1q`
| uname -m = `(uname -m) 2>/dev/null || echo unknown`
| uname -r = `(uname -r) 2>/dev/null || echo unknown`
| uname -s = `(uname -s) 2>/dev/null || echo unknown`
| uname -v = `(uname -v) 2>/dev/null || echo unknown`
|    
| /usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null || echo unknown`
| /bin/uname -X     = `(/bin/uname -X) 2>/dev/null     || echo unknown`
| 
| /bin/arch              = `(/bin/arch) 2>/dev/null              || echo 
unknown`
| /usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null       || echo 
unknown`
| /usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null || echo 
unknown`
| hostinfo               = `(hostinfo) 2>/dev/null               || echo 
unknown`
| /bin/machine           = `(/bin/machine) 2>/dev/null           || echo 
unknown`
| /usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null       || echo 
unknown`
| /bin/universe          = `(/bin/universe) 2>/dev/null          || echo 
unknown`
| 
| PATH = $PATH
| 
| _ASUNAME
| } >&5
| 
| ======================================================================
| 
| Has the autoconf team seen this already?  I ask here to shorten the
| time for finding the problem...

Thanks!

I'd like to make sure that the problem is really wrt the number, and
not wrt the structure of the command itself.

Could that person try a simple:

~/src/tiger % cat /tmp/foo.sh                                    nostromo 18:09
#! /bin/sh

exec 5> /tmp/foo.log
echo foo >&5
cat /tmp/foo.log
~/src/tiger % sh -x /tmp/foo.sh                                  nostromo 18:09
+ exec
+ echo foo
+ cat /tmp/foo.log
foo


If this fails, I'd like to know the reason of the failure.  We are in
a deep s*t, IMHO.  Because it means we have to move to another fd,
which means... that you want us to move to a fd, say 7, which we
*never* tried to use before.  In other words, to please this machine,
you're asking us to take risks on possibly many others :(

And I don't even want to imagine the penalty of having to `>>
config.log' instead of `>&5' :(

F*k.

I really should install spies in Autoconf's test suite to browse the
FDs that are available.