Re: Suspicious warning in W64 build

[Eli Zaretskii]

> From: Richard Copley <address@hidden>
> Date: Thu, 7 Sep 2017 21:02:05 +0100
> Cc: Eli Zaretskii <address@hidden>, Angelo Graziosi <address@hidden>, 
>       Emacs Development <address@hidden>
> 
>  On 7 September 2017 at 20:26, Paul Eggert <address@hidden> wrote:
> > Try putting an 'eassume (0 <= nglyphs);' before the line in question.
> 
> The warning goes away.

OK, I will commit that change, then.

> By the way, there's another warning which is both true and easy to fix.
> "(1 << ((n) % 32)))" should read "(1u << ((n) % 32)))" on line 2191.
> 
>   CC       w32font.o
> w32font.c: In function 'font_supported_scripts':
> w32font.c:2191:32: warning: result of '1 << 31' requires 33 bits to
> represent, but 'int' only has 32 bits [-Wshift-overflow=]
>    if (subranges[(n) / 32] & (1 << ((n) % 32))) \
>                                 ^
> w32font.c:2246:3: note: in expansion of macro 'SUBRANGE'
>    SUBRANGE (31, Qsymbol);
>    ^~~~~~~~
> w32font.c:2191:32: warning: result of '1 << 31' requires 33 bits to
> represent, but 'int' only has 32 bits [-Wshift-overflow=]
>    if (subranges[(n) / 32] & (1 << ((n) % 32))) \
>                                 ^
> w32font.c:2308:3: note: in expansion of macro 'SUBRANGE'
>    SUBRANGE (95, Qtai_le);
>    ^~~~~~~~

Yes, I've seen this and have a fix for it.  Thanks.