On 2025-04-05  08:46, Eli Zaretskii wrote:

>>> Why does this matter?  Your patch doesn't eliminate initial_argv, so
>>> initial_argv[0] still holds the name of the Emacs program, albeit only
>>> in the form it was provided by the calling shell.  Right?  If so, then
>>> for reporting errors (and ones that should never happen at that), I
>>> don't see any need to jump through hoops.
>>
>> But my patch, at least the second one [1] on message #20
>> of this bug, plus your earlier request to #ifdef away
>> function and call to copy_raw_args on Windows, would skip
>> the initialization of initial_argv on Windows, no?
> 
> Ah, I've missed that part.  Then the call to 'error' could just use
> argv[0] instead of initial_argv[0], right?

Then what do you think about the attached patch?