1. You might want to write a separate program to clean out any stale /tmp entries instead of leaving this up to script A. This way the user can reload the page and still get output, without you having to process all over again.
2. Your script B should not write directly to the /tmp file, but should create a preliminary version which it writes to, then renames it at the end to the proper name. This will prevent script A from jumping the gun and printing when script B has only written half its output.

hmm, well the thing is that script A does not need anything at all from script B to execute. Most of the time, script A wont even call script B, but when it does, it does not have to wait for script B to complete, but it does. I think I need to be more clear.. script A simply prints the webpage, and also checks to see if a certain database file is too old. If that file is old, script A calls script B (currently using server(scriptB.cgi);) which downloads a fresh database file from a remote site (this takes a couple minutes cause its a big file). Script A COULD use the old db file until script B is done, but it doesnt because it waits for the server call to complete.

In that case, don't bother calling script B a ".cgi", since it is really a stand-alone process. CGI programs are intended to be run directly by the client, which means they operate a little differently than regular programs. For instance, using CGI.pm, they require a bit of input from STDIN to get going, unless you specify parameters on the command line.

How about this:

if (file_is_too_old())
{
     unless (fork())
     {
           exec ("update_script");
     }
}
[download]

Remember that, unless you're using an SUID script, the "update_script" will be run as your Web user, which is traditionally 'nobody'. Make sure that this user has all the required privileges to update your files, or that the SUID user does.

There was a really good discussion of doing this sort of thing here Searching large files before browser timeout which includes (amongst other things) a node I wrote one way to do this and also a link to one of merlyn's columns on the same subject which covers the topic in detail.

cheers

tachyon

s&&rsenoyhcatreve&&&s&n.+t&"$'$`$\"$\&"&ee&&y&srve&&d&&print

# print cgi response as usual, then...

my $pid;

if ($pid = fork) {
  # display run, so exit
  exit(0);
}
elsif (defined $pid) {
  # run script b function if $run_script_b true...
  if ($run_script_b) {
    script_B();
  }
}
else {
  # raise an error coz the fork failed
  ...
}
[download]

Works for me. My script starts an FTP install that can take up to half an hour to run, depending on network connection.

cLive ;-)