Re^2: Inappropriate ioctl for device error on system call

The $fullfilename exists. There are some commands that run earlier that work fine:


# Unzip file here
system("gunzip $fullfilename") == 0 
   or die "Cannot gunzip $fullfilename $!";

open FH, "file $fullfilename | " 
   or die "Cannot open fullfilename: $fullfilename $!";
my $filetype = <FH>;
close FH;

if ($filetype =~ /ASCII/ || $filetype =~ /ISO-8859/) {
   latex_template("$dirname/print.files.txt", $filename, $data[2]);
   system("latex banner.$$.tex") == 0 
      or die "Cannot latex banner.$$.tex $!";
   system("dvips banner.$$.dvi -o banner.$$.ps") == 0 
      or die "Cannot dvips banner.$$.dvi $!";

   if ( -e $fullfilename ){
      print "File to be enscripted ($fullfilename) exists";
   }

   system("$ENSCRIPT $fullfilename -o $TEMPDIR/$dirname.$filename.ps")
+ == 0 
      or die "Cannot $ENSCRIPT $fullfilename -o $TEMPDIR/$dirname.$fil
+ename.ps: $!";
        
   system("lp -d $PRINTER_NAME banner.$$.ps 2>$TEMPDIR/error.txt") == 
+0 
      or die "Cannot lp -d banner.$$.ps $!";
        
   system("lp -d $PRINTER_NAME -n $num_copies -q $job_priority $page_l
+ist $TEMPDIR/$dirname.$filename.ps 2>$TEMPDIR/error.txt") == 0 
      or die "Cannot lp -d $TEMPDIR/$dirname.$filename.ps $!";
[download]

I also tried replace the output file ($TEMPDIR/$dirname.$filename.ps) with a known good location. The enscript command actually does create the postscript file. This makes the existence of the error that much more confusing.

Neil Watson
watson-wilson.ca

Comment on Re^2: Inappropriate ioctl for device error on system call Select or Download Code

Replies are listed 'Best First'.
Re^3: Inappropriate ioctl for device error on system call by traveler (Parson) on Dec 12, 2006 at 19:39 UTC
Some more troubleshooting hints: Does it run from the command line? If so, perl is probably not giving expected access to some filehandle. Try tracing it with strace. You can do it from the command line if the command version fails, or from perl by adding it to the system call. It can generate a lot of output but may show what file descriptor is causing the problem.	[reply]
Re^4: Inappropriate ioctl for device error on system call by neilwatson (Priest) on Dec 12, 2006 at 20:19 UTC
The enscript command does work from the command line and in the perl script. For some reason it seems that the `== 0 or die` section evaluates to the die. I do not know why considering that the other system calls work without dying. Also, I checked the `$!` variable after earlier system calls and the contents was already "Inappropriate ioctl for device" so it seems that this message may not be related. Neil Watson watson-wilson.ca	[reply] [d/l] [select]
Re^5: Inappropriate ioctl for device error on system call by traveler (Parson) on Dec 12, 2006 at 21:12 UTC
In that case note that enscript may have had a non-fatal termination: `Enscript returns value 1 to the shell if any errors were encountere +d. On successful termination, the return code is constructed from +the fol- lowing flags: 0 no errors or warnings 2 some lines were truncated or wrapped 4 some characters were missing from the used fonts 8 some characters were unprintable` [download] Note the On successful termination: Maybe some lines were truncated or some characters unprintable.	[reply] [d/l]