Parent/child choice in sslio

I've made use of sslio to add TLS support to Bruce Guenter's mailfront 
(see the bgware list for details). But I found debugging quite confusion 
because of the decision that sslio makes about the allocation of tasks to 
the parent and child after it forks.

I had modified smtpfront-qmail so that it would accept and acknowledge a 
"STARTTLS" command, then exec "sslio ... smtpfront-qmail".

What confused me was than I had an smtpfront-qmail process (pid N) before 
I gave the STARTTLS command, then had an smtpfront-qmail process (pid N) 
and child sslio process (pid N+1) after. It took a bit of head scratching 
to realise that process N was smtpfront-qmail, then sslio, then 
smtpfront-qmail again.

This patch would have avoided the confusion:

--- sslio.c.orig	Thu May 13 13:46:54 2004
+++ sslio.c	Thu May 13 13:47:38 2004
@@ -286,7 +286,7 @@
   if (pipe(encpipe) == -1) fatal("unable to create pipe for encoding");
   if (pipe(decpipe) == -1) fatal("unable to create pipe for decoding");
   if ((pid =fork()) == -1) fatal("unable to fork");
-  if (pid == 0) {
+  if (pid) {
     if (close(encpipe[1]) == -1)
       fatal("unable to close encoding pipe output");
     if (close(decpipe[0]) == -1)

I realise that you might want to wait for the child to terminate, and pass 
its status back as exit status of sslio, but I don't need that now.

---
Charlie