Re: master f4af1137e6: Bind print-symbols-bare to t in byte-compile-from-buffer

Re: master f4af1137e6: Bind print-symbols-bare to t in byte-compile-from-buffer

[Stefan Monnier]

> -  (let ((print-symbols-bare t))
> +  (let ((print-symbols-bare t))         ; Possibly redundant binding.
>      (macroexpand `(declare-function ,fn ,file ,@args))))

Why do we need this binding here?
This is just a `macroexpand`, so it's not expected to print anything
at all.  What kind of prints are supposed to be affected, and if so, why
should it be solved by a let-binding here rather than elsewhere, such
as maybe in `macroexpand`?


        Stefan

Re: master f4af1137e6: Bind print-symbols-bare to t in byte-compile-from-buffer

[Alan Mackenzie]

Hello, Stefan

On Sun, Jan 23, 2022 at 11:47:34 -0500, Stefan Monnier wrote:
> > -  (let ((print-symbols-bare t))
> > +  (let ((print-symbols-bare t))         ; Possibly redundant binding.
> >      (macroexpand `(declare-function ,fn ,file ,@args))))

> Why do we need this binding here?
> This is just a `macroexpand`, so it's not expected to print anything
> at all.  What kind of prints are supposed to be affected, and if so, why
> should it be solved by a let-binding here rather than elsewhere, such
> as maybe in `macroexpand`?

I don't think we do need that binding there.  The main point of my patch
was to bind the variable at a high level (in byte-compile-from-file),
and I don't think we need it anywhere else (execept compile-defun).

But I didn't have the courage to strip it out, yesterday.  Maybe I
should have done.

>         Stefan

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: master f4af1137e6: Bind print-symbols-bare to t in byte-compile-from-buffer

[Stefan Monnier]

Alan Mackenzie [2022-01-24 11:50:36] wrote:
> Hello, Stefan
> On Sun, Jan 23, 2022 at 11:47:34 -0500, Stefan Monnier wrote:
>> > -  (let ((print-symbols-bare t))
>> > +  (let ((print-symbols-bare t))         ; Possibly redundant binding.
>> >      (macroexpand `(declare-function ,fn ,file ,@args))))
>
>> Why do we need this binding here?
>> This is just a `macroexpand`, so it's not expected to print anything
>> at all.  What kind of prints are supposed to be affected, and if so, why
>> should it be solved by a let-binding here rather than elsewhere, such
>> as maybe in `macroexpand`?
>
> I don't think we do need that binding there.  The main point of my patch
> was to bind the variable at a high level (in byte-compile-from-file),
> and I don't think we need it anywhere else (execept compile-defun).

Just to clarify:

My question was not "Is there a path to here that doesn't already bind
that var to t?" (to which you already reply in the comment), but "(why) do
we need to have that variable be t during the execution of that code?".


        Stefan