bug#64537: mix lists and vectors when using backquote

bug#64537: mix lists and vectors when using backquote

[Shynur Xie]

The docstring of ‘backquote’ says “vectors work just like lists”.
Should we also clarify that mixing lists and vectors these 2 way:

    `(... ,@[...])  ; by contrast: `(,@[] 1)  => (1)
    `[,@'()]        ; by contrast: `[1 ,@'()] => [1]

will lead to unexpected behavior?  Or suggest not doing so?  Or let
‘backquote’ detect this and signal an error?

BTW, if [[https://debbugs.gnu.org/cgi/bugreport.cgi?bug=64415]] is
actually not a bug, please consider document that behavior.  Thanks.

bug#64537: mix lists and vectors when using backquote

[Stefan Kangas]

Shynur Xie <one.last.kiss@outlook.com> writes:

> The docstring of ‘backquote’ says “vectors work just like lists”.
> Should we also clarify that mixing lists and vectors these 2 way:
>
>     `(... ,@[...])  ; by contrast: `(,@[] 1)  => (1)
>     `[,@'()]        ; by contrast: `[1 ,@'()] => [1]
>
> will lead to unexpected behavior?  Or suggest not doing so?  Or let
> ‘backquote’ detect this and signal an error?

Copying in Stefan Monnier.

bug#64537: mix lists and vectors when using backquote

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

>> The docstring of ‘backquote’ says “vectors work just like lists”.
>> Should we also clarify that mixing lists and vectors these 2 way:
>>
>>     `(... ,@[...])  ; by contrast: `(,@[] 1)  => (1)
>>     `[,@'()]        ; by contrast: `[1 ,@'()] => [1]
>>
>> will lead to unexpected behavior?  Or suggest not doing so?  Or let
>> ‘backquote’ detect this and signal an error?
>
> Copying in Stefan Monnier.

Is there a good reason why we have those unexpected behaviors?


        Stefan

bug#64537: mix lists and vectors when using backquote

[Stefan Kangas]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>> The docstring of ‘backquote’ says “vectors work just like lists”.
>>> Should we also clarify that mixing lists and vectors these 2 way:
>>>
>>>     `(... ,@[...])  ; by contrast: `(,@[] 1)  => (1)
>>>     `[,@'()]        ; by contrast: `[1 ,@'()] => [1]
>>>
>>> will lead to unexpected behavior?  Or suggest not doing so?  Or let
>>> ‘backquote’ detect this and signal an error?
>>
>> Copying in Stefan Monnier.
>
> Is there a good reason why we have those unexpected behaviors?

I see two unexpected behaviors:

  1.   `(... ,@[...])  =>  (\... . [\...])
  2.   `[,@'()]        =>  [(\,@ 'nil)]

I'd say these are the expected values:

  1.   `(... ,@[...])  =>  (\... \...)
  2.   `[,@'()]        =>  []

I don't see any good reason for them, so perhaps these are just plain
bugs to be fixed?

bug#64537: mix lists and vectors when using backquote

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

>     `(... ,@[...])  ; by contrast: `(,@[] 1)  => (1)

AFAICT this is the result of an optimization:

    `(BAR ,@FOO)

is optimized to

    `(BAR . ,FOO)

I expect a fair bit of code "relies" on this optimization, so I'm not
sure we want to "fix" it.

>     `[,@'()]        ; by contrast: `[1 ,@'()] => [1]

This looks like a plain bug (probably linked to another optimization).

    `[,@(not t)]

doesn't suffer from this problem but

    `[,@'[]]

also suffers from it.


        Stefan