Re: Building Emacs on MSYS2

[Óscar Fuentes]

Eli Zaretskii <address@hidden> writes:

>> >> >>    local with_wide_int='no'
>> >> >>
>> >> >>    if test "${CARCH}" == 'x86_64'; then
>> >> >>      with_wide_int='yes'
>> >> >>    fi
>> >> >
>> >> > You shouldn't need this. --with-wide-int is effective only on 32-bit
>> >> > platforms; integers are always wide on 64-bit platforms.
>> >> 
>> >> Right. Maybe the person who introduced that code on the PKGBUILD worried
>> >> about the possibility of changing the default on x86 to wide-int. IIRC
>> >> there was some discussion about that on the past.
>> >
>> > The discussions were about the 32-bit build.  If anything, the script
>> > should use --with-wide-int in the 32-bit build, i.e. exactly the
>> > opposite of what it does now.
>> 
>> My point is that the PKGBUILD author possibly was "protecting" the
>> package against upstream changing the x86 default to --with-wide-int.
>
> It does exactly the opposite, and to the non-x86 build, so I don't
> understand how did you reach that conclusion.

You misread the code. It unconditionally sets with_wide_int='no' and
*then* checks if the arch is x86_64, in which case sets
with_wide_int='yes'.

Hence the code is explicitly setting --with-wide-int on both cases to a
value which happens to coincide with the default *for now*.