[PATCH] xdiff: fix trivial build warnings on Windows

[Matt Harbison]

On Sun, 04 Mar 2018 16:45:28 -0500, Matt Harbison <mharbison72 at gmail.com>  
wrote:

> # HG changeset patch
> # User Matt Harbison <matt_harbison at yahoo.com>
> # Date 1520197662 18000
> #      Sun Mar 04 16:07:42 2018 -0500
> # Node ID c4a6b599a46f93070f5492c9e68566e6be570d2f
> # Parent  1f9bbd1d6b8ae4f7ea5d9f4310269a3b0242e7b0
> xdiff: fix trivial build warnings on Windows
>
> These are mostly size_t to int/long conversions that are obviously safe,  
> along
> with a signed/unsigned comparison.  I don't have clang, so I tried  
> following the
> existing whitespace convention in each module.

The remaining xdiff warnings are:

mercurial/thirdparty/xdiff/xmerge.c(352) : warning C4244: '=' : conversion  
 from '__int64' to 'long', possible loss of data
mercurial/thirdparty/xdiff/xmerge.c(355) : warning C4244: '=' : conversion  
 from '__int64' to 'long', possible loss of data
mercurial/thirdparty/xdiff/xutils.c(412) : warning C4244: '=' : conversion  
 from '__int64' to 'long', possible loss of data
mercurial/thirdparty/xdiff/xutils.c(415) : warning C4244: '=' : conversion  
 from '__int64' to 'long', possible loss of data

These are a bit more concerning, because it looks like two randomish  
pointers are subtracted.  I haven't spent much time trying to understand  
the code, so presumably there's something ensuring that these values stay  
close to each other?  'long' is only 32 bit on Windows, so maybe the size  
field in mmbuffer_t and mmfile_t should be size_t/ptrdiff_t?