<method try
  if_error    =null=wob=ekind.expression
  if_timed_out="timed_out"=wob=ekind.expression
  finally     =opt=wob=ekind.expression
  time_limit  =null
  _other_unkeyed=opt=wob=ekind.expression="_body"/>

<try if_error="busted">
   123
   456
</try>

456

<try if_error="we had an error">
    234
    <error "busted"/> 
    567 
   </try>

"we had an error"

<try if_error="there was an error" > 234 567 </try>

567

a wob When the if_error argument of a try is executed, a local variable named error_object is bound to the string of the error message that was created by executing the attributes to try. If the body of a call to 'try' causes an error, try executes the code up to the error in the body, and, with no if_error argument, just returns null without erroring. Since the call to try above doesn't actally error, any code after the call to try will be executed. If you want to return the "error_object" (which is just string), you can: Beware, though, that the call to try does not result in an error, and processing continues on after the call to try: If what you're after is to call the try and have it error when its content errors: The above code is the same as: So you wouldn't bother using the try at all. More likely though, is that you'd want to do something before causing the error in the if_error argument. You might want to write some 'clean-up' code before erroring. Here we simulate our clean-up code with an echo statement: If you want to protect against the if_error argument of a try call causing an error, wrap another try statement all around your original call to try. busted Here's the order of execution above:

1. 123 is executed
2. <error "a bug"/> is executed causing an error.
3. <error "some code that errors"/> is executed
4. "busted" is executed.

<try time_limit=20 if_timed_out="timed out">
   <for_each forever=true> null </>
  </try>

"timed out"