Re: Building Diffutils with MinGW

[Eli Zaretskii]

> Date: Fri, 04 May 2012 14:28:30 -0700
> From: Paul Eggert <address@hidden>
> CC: address@hidden
> 
> On 05/04/2012 01:52 PM, Eli Zaretskii wrote:
> > That's okay, but then the configure script will have to add the
> > "-I../ms" switch to the GCC command-line arguments.
> 
> That'd be fine.  I assume that there will be a file ms/configure
> that does the right thing; perhaps it could be a thin wrapper
> around regular 'configure', or perhaps it'll need to be built
> some other way.

I don't understand why should something like that be done by a
separate script, since the top-level configure already includes
multiple tests for mingw* configurations.

Another issue: This fragment of src/system.h:

  #include <dirent.h>
  #ifndef _D_EXACT_NAMLEN
  # define _D_EXACT_NAMLEN(dp) strlen ((dp)->d_name)
  #endif

is sub-optimal for systems that have the d_namlen member in struct
dirent, but don't define _D_EXACT_NAMLEN.  And since the configure
script doesn't have a test for d_namlen, there's no HAVE_* macro to
modify the above snippet to take such systems into consideration.  Is
the d_namlen member ignored on purpose?