On Fri, 20 Nov 2026, "Dr. Arne Babenhauserheide" <[email protected]> wrote:
> Olivier Dion <[email protected]> writes:
>
>> On Thu, 19 Nov 2026, Janneke Nieuwenhuizen <[email protected]> wrote:
>>>> GNU Guile 3.0.12, release candidate 1 is now available.
>>>
>>> That's an impressive list of changes, thanks!  One quick question after
>>> merely compiling two of our projects.
>
>> You can still use them.  However, warnings are emitted on
>> `current-warning-port'.
>>
>>> --8<---------------cut here---------------start------------->8---
>>> (let ((f' (transform f t))      ;lots of warnings with guile-3.0.12rc1
>>>       (f'' (transform f' t)))   ;likewise; see below
>>>   (calc f f' f'') ...)
>>> --8<---------------cut here---------------end--------------->8---
>>>
>>> now gives a lot of warnings.  Is this intentional, and if so, do we
>>> really want to disallow this, and why?  Are we the only ones using this?
>>> Thoughts?
>>
>> The warnings are meant to tell you that we will remove this read
>> behavior in the next breaking release.  Probably in 3.2.
>>
>> The rationale was to be closer to what other Scheme do, e.g. the quote
>> character is a toke separator.  See the full discussion here:
>> https://codeberg.org/guile/guile/issues/223
>
> With the new version, I now also see a lot of warnings for dryads-wake
> that uses quoted prose (I just checked). For example this now throws
> warnings, so Guile 3.2 would break it:
>
> (define (main args)
>   (Enter (First Witch)
>           (Second Witch)
>           (Third Witch)
>           (First Eldritch))
>
>   (Second Witch :resolute
>       (When the hurlyburly's done,)
>       (When the battle's lost and won.)))
>
> The lines below Second Witch are implicitly quasiquoted.
>
> Losing that would hurt a lot and I’m not sure whether I could change it
> (and usability of enter-three-witches would be permanently reduced,
> because using ' within a word is just what people do intuitively).
>
> Though f' is the much more generally visible example of problems the
> change causes (because representing the derivatives as f' or f'' is
> extremely elegant).
>
> I’m glad that the warning made these problems visible.
>
> And I’d really prefer not to see this broken.

I think we can have a knob for the reader to enable this as an
extension.  However, for the sake of portability, I believe this should
be opt-in instead of the default.

What do you think?  You would need to do add something like:

#!quote-in-token

at the begining of your Scheme file.

-- 
Olivier Dion

Reply via email to