Re: QUIT and ABORT

On 2025-05-24 17:54, dxf wrote:

On 24/05/2025 7:20 pm, Ruvim wrote:

On 2025-05-20 09:49, dxf wrote:

On 19/05/2025 8:52 pm, Ruvim wrote:

On 2025-05-18 18:27, dxf wrote:

On 18/05/2025 6:16 pm, Ruvim wrote:

...
Note that in `error` you don't transfer control to `(abort)` only in the case of `-56`.

>
I transfer control to (abort) except in the case of -56 where it passes to (quit).

>
Hence, `-56` is a special case in your implementation.

>
No more special than what ANS did for -1 and -2 .

>
Anyway, ANS does not define any special *behavior* for `-56`, does it?

 But you agree the special behaviour is necessary for a system with
catchable QUIT where QUIT is executed but no exception frame exists?

It is a possible option, but it is not necessary.  Since technically a catchable `quit` can be specified without special behavior of `throw` for `-57`.
For example: if there is an exception frame, perform `-57 throw`, otherwise perform the function of CORE `quit`. But this complicates the system's implementation without any profit.

Reasons to not generally define a special behaviour for -56 ...
a) only systems that want a catchable QUIT need it
b) not every system is capable of providing the special behaviour

It seems you mean that different systems may behave differently in this regard. Then it must be *explicitly* specified as implementation-defined option, or as an ambiguous condition.
"If you want to leave something open to the system implementor, make that explicit, e.g., by making it an ambiguous condition"
<https://forth-standard.org/standard/process>

[...]
>

You interpret the reservation of `-56` for `quit` as a provision to make `quit` catchable.

>
Correct.  I'm grateful for serious opponents as they do my homework for me.
If I had any concerns about a catchable QUIT beforehand, your posts have
resolved them.

>

I've seen no convincing rationale that QUIT should not have
the same entitlement as ABORT and ABORT" .

>
This train has long since left.
>

[...]

 Just as ANS allowed ABORT and ABORT" to be catchable or not and the application
programmer had to factor that in, so must he do for QUIT.

To allow a catchable `quit`, ANS should have added a version of `quit` in the Exception word set  (similar to `abort` and `abort"`).
[...]

 It's unlikely a 'Request for Clarification' is applicable as ANS TC no longer
exists.  That leaves making a proposal reflecting the outcome that you want.

So the only way to allow a catchable `quit` is to make a proposal that specifies a version of `quit` in the Exception word set (and changes `throw`, if any).
--
Ruvim