Re: rm: don't waste time preprocessing on tmpfs or nfs

[Eric Blake]

Jim Meyering <jim <at> meyering.net> writes:

> 
> I'll probably be using this change:
> 
> From ebb424314a577c0b8ccaf3ef79c3c9a3379b31a5 Mon Sep 17 00:00:00 2001
> From: Jim Meyering <meyering <at> redhat.com>
> Date: Wed, 1 Oct 2008 12:21:58 +0200
> Subject: [PATCH] remove.c: combine two helper functions, to avoid sign-
extension

Even with that, the build is still failing on cygwin:

remove.c: In function `dirent_inode_sort_may_be_useful':
remove.c:1306: error: `S_MAGIC_TMPFS' undeclared (first use in this function)
remove.c:1306: error: (Each undeclared identifier is reported only once
remove.c:1306: error: for each function it appears in.)
remove.c:1307: error: `S_MAGIC_NFS' undeclared (first use in this function)
remove.c: At top level:
remove.c:1290: warning: 'dirent_inode_sort_may_be_useful' defined but not used

Since we've already applied 58aaf71 to silence other functions when d_type is 
unavailable (and sorting is useless), it seems like that needs to be repeated 
here.  The constants from "fs.h" are not defined, since that file is valid only 
for Linux and Hurd, but commenting out the entire function bypasses that 
problem.

-- 
Eric Blake