Re: [Emacs-diffs] master 5f72004: Revert "Fix command repetition with lexical-binding (Bug#29334)"

[Noam Postavsky]

On Sun, Jan 7, 2018 at 10:36 AM, Stefan Monnier
<address@hidden> wrote:
>> --- a/src/callint.c
>> +++ b/src/callint.c
>> @@ -357,9 +357,7 @@ invoke it.  If KEYS is omitted or nil, the return value 
>> of
>>        /* Compute the arg values using the user's expression.  */
>>        specs = Feval (specs,
>>                    CONSP (funval) && EQ (Qclosure, XCAR (funval))
>> -                     ? CAR_SAFE (XCDR (funval))
>> -                     : COMPILEDP (funval) && INTEGERP (AREF (funval, 
>> COMPILED_ARGLIST))
>> -                     ? Qt : Qnil);
>> +                  ? CAR_SAFE (XCDR (funval)) : Qnil);

> Why?  This looks wrong.

Hmm, I just reverted to the previous code without actually thinking
about it. But after checking this more closely, I find that passing
non-nil LEXICAL to `eval' doesn't really do anything for compiled
functions anyway: the "lexicalness" has already been applied when
compiling. E.g., the above change has no effect on the attached tests.

callint-tests.el
Description: Binary data