Re: Unable to build devel/libuv on NetBSD 6_1_Stable

[Joerg Sonnenberger <joerg%bec.de@localhost>]

On Mon, Jan 23, 2017 at 09:51:33PM -0500, Ian D. Leroux wrote:
> [sorry for duplicate, failed to cc list the first time]
> 
> On Mon, 23 Jan 2017 13:51:25 -0000 yancm%SDF.ORG@localhost wrote:
> > I'm still having problems with this package, has this caused anyone
> > else problems? Has anyone updated and built successfully on NetBSD 6x
> > (or 7x)?
> 
> I just rebuilt libuv-1.10.2 on NetBSD/amd64 7.99.29 with no problems.
> 
> > src/unix/fs.c: In function 'uv__to_stat':
> > src/unix/fs.c:825:28: error: 'struct stat' has no member named 'st_atim'
> > src/unix/fs.c:826:29: error: 'struct stat' has no member named 'st_atim'
> > src/unix/fs.c:827:28: error: 'struct stat' has no member named 'st_mtim'
> > src/unix/fs.c:828:29: error: 'struct stat' has no member named 'st_mtim'
> > src/unix/fs.c:829:28: error: 'struct stat' has no member named 'st_ctim'
> > src/unix/fs.c:830:29: error: 'struct stat' has no member named 'st_ctim'
> > src/unix/fs.c:833:32: error: 'struct stat' has no member named 'st_birthtim'
> > src/unix/fs.c:834:33: error: 'struct stat' has no member named 'st_birthtim'
> > src/unix/fs.c: In function 'uv__fs_done':
> 
> It looks like there's some twisty #ifdefs in fs.c that decide whether
> those field names get an 'e' at the end or not. According to stat(2),
> on NetBSD they should have an 'e'. I'm not actually sure why this is
> working for me, since it looks to me like the #ifdefs get this wrong
> (e.g. at line 832 in fs.c). Someone with a better knowledge of stat(2)
> and its (evidently inconsistent) behaviour across platforms should
> chime in.

The version without 'e' is the POSIX 2008 spelling. That's simply not
supported on NetBSD 6.

Joerg