From: dxforth@gmail.com   
      
   On 8/05/2025 10:50 pm, Ruvim wrote:   
   > On 2025-05-08 06:52, dxf wrote:   
   >> On 8/05/2025 3:52 am, Ruvim wrote:   
   >>> ...   
   >>> I use `quit` only for testing and debugging. And I expect to get into the   
   Forth text interpreter (Forth shell) exactly in the current program state (as   
   far as possible).   
   >>>   
   >>> If `quit` simply calls `-56 throw`, the program state will probably change   
   (due to actions after `catch` in the program, including restoring the data   
   stack depth in `throw`) when you get into the Forth shell. Also, there is a   
   chance that you will not    
   get into the Forth shell at all if the program does not re-throw the error in   
   some places.   
   >>>   
   >>> So I would not recommend the suggested deviation in the `quit` behavior   
   even for new implementers, since this brakes the well-known expectation from   
   `quit`. A better way is to introduce another word with desired behavior   
   deviation.   
   >>   
   >> ANS never gave a rationale for ABORT ABORT" QUIT - only options that we're   
   now   
   >> exploring.   
   >>   
   >> For the reasons you've stated I'm reluctant to define:   
   >>   
   >> : QUIT -56 THROW ;   
   >>   
   >> But are there reasons an application might were THROW able to handle it?   
   >   
   > I don't see such a reason. The application can simply do `abort` or `-56   
   throw` (and no special cases are needed).   
      
   Yes but those would be workarounds.  It's making the argument QUIT can't or   
   shouldn't   
   be caught whereas I'm talking about entitlement.  Speaking of entitlement, ANS   
   made   
   EXCEPTION EXT a one-way street.  There's no rolling back from the changes.   
      
   > If you need the Forth shell in the application data stack and state — do   
   `quit`. If you need to break the application execution — do `abort`.   
   >   
   >   
   > Probably, the external signal SIGTERM can be transferred to `-56 throw` in   
   the Forth system, and the external signal SIGINT (Ctrl+C) can be transferred   
   to `-28 throw`.   
   >   
   >   
   >   
   >> It's arguable ANS left the door open. Just because the spec for THROW didn't   
   >> special-case -56 doesn't mean an implementer can't.   
   >   
   > The implementer can special-case -56 (as well as anything else) only to the   
   extent that it does not affect a standard program.   
   >   
   >   
   > --   
   > Ruvim   
   >   
      
   --- SoupGate-DOS v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)