Pitfalls

• The result of a subroutine call cannot be looked up directly in a database map. Consider this RHS of a rule:
  
  

  $( uucp $>96 $1 $)
  

  Here, the intention is to pass $1 to rule set 96 and then to look up the result in the uucp map. Instead, the literal value 96 and the value in $1 are looked up together and fail first. Then $1 is passed to rule set 96, and the result of that subroutine call becomes the result of the RHS.

• The %s of IDA sendmail databases will not work as is with V8 sendmail. To make it work, replace it with a %1 throughout the IDA source file, then reverse the keys and values, and finally recreate the database file. This conversion can easily be automated with sed(1) or awk(1).
• If you are running a Solaris 2.4 or earlier release of Sun's operating system, your database files should not live on tmpfs mounted file systems. File locking was not implemented for tmpfs until Solaris 2.5.
• Avoid assuming that all K command switches mean the same thing for all classes. The ad hoc nature of class submissions by outsiders makes that assumption perilous.
• Not all initialization errors or lookup errors are reported. For some of them you will see an indication of an error only if you use the -d38.2 debugging switch (see -d38.2).