Re: Towards a reasonable unwinding flow-control semantics.

At 07:06 PM 8/16/00 -0400, Chaim Frenkel wrote:
> >>>>> "GB" == Graham Barr <[email protected]> writes:
>
> >> There is one case to be considered, what if the try block wishes
> >> to avoid its own catch clauses, and start the unwinding with the
> >> uplevel try block.
>
>GB> Can you think of a time you would want that ?
>
>Here's one. If the catch blocks translate exceptions into a more
>domain specific exception. And the try block already knows what to do
>with the error. No need to put in a special case, just let it leak
>past the catches.
>
>But more directly, if the exception is directly detected/raised in the
>try block itself. I probably wouldn't want to route it through the
>local catches.
>
>The try block knows what the situation is, it did it. So the exception
>is aimed at the caller not at itself.

I think this has merit.  We want throw to be able to throw farther.  Maybe:

         throw_up Exception::Foo(...);

to signify that we're going up instead of out.

Okay, :-)  Got a better syntax?  Would you want the option of throwing up n 
(> 1) levels?

--
Peter Scott
Pacific Systems Design Technologies