Re: Suspicious warning in W64 build

[Eli Zaretskii <eliz@gnu.org>]

> From: Richard Copley <rcopley@gmail.com>
> Date: Thu, 14 Sep 2017 20:34:42 +0100
> Cc: Fabrice Popineau <fabrice.popineau@centralesupelec.fr>, 
> 	Emacs Development <emacs-devel@gnu.org>
> 
> Not much. Here's another transcript. It's a lot better.

Thanks, please try the latest master.  I hope I fixed most of the
warnings in Windows specific code there.

I must say that GCC 7 static code analysis is really imperfect, not
good enough for producing useful warnings from its results, because
some of the warnings are outright wrong.  For example this one:

  w32term.c: In function 'w32_read_socket':
  w32term.c:4979:10: warning: 'button' is used uninitialized in this function [-Wuninitialized]
	int button;
	    ^~~~~~
  w32term.c:3099:16: warning: 'button' may be used uninitialized in this function [-Wmaybe-uninitialized]
     result->code = button;
     ~~~~~~~~~~~~~^~~~~~~~
  w32term.c:3090:7: note: 'button' was declared here
     int button;
	 ^~~~~~
  w32term.c:3104:6: warning: 'up' may be used uninitialized in this function [-Wmaybe-uninitialized]
	    | (up
	      ~~~
	? up_modifier
	~~~~~~~~~~~~~
	: down_modifier));
	^~~~~~~~~~~~~~~~
  w32term.c:3091:7: note: 'up' was declared here
     int up;
	 ^~
  w32term.c:5018:9: warning: 'up' may be used uninitialized in this function [-Wmaybe-uninitialized]
	if (up)
	   ^

Evidently, GCC 7 doesn't see that the function parse_button
initializes these 2 variables.

About this warning:

  w32.c:7550:1: warning: no previous prototype for 'sys_strerror' [-Wmissing-prototypes]
   sys_strerror (int error_no)
   ^~~~~~~~~~~~

What MinGW64 header has the prototype of strerror?  I thought it was
string.h, like in mingw.org's headers, but evidently that's not true?
If the prototype is in string.h, then why is this warning being
displayed?

A few warnings are not specific to Windows, I will describe them in a
separate message.

Thanks for the transcript, it was really useful.