Thank you for your feedback, John.

Am Mo., 30. Okt. 2023 um 01:10 Uhr schrieb John Cowan <[email protected]>:

> Comments:
>
> 1. I propose the name with-continuation-handler, which is not present in
> SRFI 226.
>

As we discussed on #scheme, the name will be `with-unwind-handler'.


> 2. In the first paragraph of the Rationale, read "supplemented" instead of
> "replaced".  Call/cc will not be replaced.
>

Fixed.


> 3. The Examples section should be moved down to after the Specification
> section.
>

I think there are arguments in favour and against; I usually like to see
examples (even if I cannot understand them fully at that time) before
diving into the specification.  If there are strong voices to move this
subsection down, I am going to do it.


> 4. The syntax of `guard` should be properly indented.
>

I don't know how to do it (especially while keeping the font choices right
- some are monospaced, others are not), so I am not taking any action
here.  `Guard' is not improperly indented, it just occurs on a single line.

5.  Dynamic-wind will have to be redefined in an implementation that
> redefines call/cc.
>

Not necessarily so, I think. If the new call/cc calls the old call/cc, it
inherits the old call/cc's dynamic-wind as well, which may still be
appropriate.

The new text is now: "A portable implementation of delimited continuations
(which uses <code>call/cc</code> and global state) is only possible by at
least replacing the user-visible <code>call/cc</code>."

@Arthur A. Gleckler <[email protected]>: I have sent you a pull request
for the second draft, which incorporates the above.


>
>
>
> On Sun, Oct 29, 2023 at 6:44 PM Arthur A. Gleckler <[email protected]>
> wrote:
>
>> Scheme Request for Implementation 248,
>> "Minimal delimited continuations",
>> by Marc Nieper-Wißkirchen,
>> is now available for discussion.
>>
>> Its draft and an archive of the ongoing discussion are available at
>> https://srfi.schemers.org/srfi-248/.
>>
>> You can join the discussion of the draft by filling out the subscription
>> form on that page.
>>
>> You can contribute a message to the discussion by sending it to
>> [email protected].
>>
>> Here's the abstract:
>>
>> Scheme's exception system is extended so that exception handlers gain
>> access to the delimited continuation representing the rest of the
>> computation of the call of the thunk guarded by the handler. Algebraic
>> effect handlers can be directly expressed in this extended exception
>> system. The system directly implements the shift0/reset0 delimited
>> control operators. It is well known that other delimited control operators
>> like prompt0/control0 or reset/shift are expressible in shift0/reset0
>> (and vice versa).
>>
>> Regards,
>>
>> SRFI Editor
>>
>

Reply via email to