On 07/18/2018 08:33 AM, Barath Aron wrote:
In order to test things, I'll install a NetBSD and (probably) a
FreeBSD. Despite I have ideas, I want to install these systems before
changing/suggesting anything.
I didn't have enough time to install them, but I looked into the gnulib
itself.
I made a clone, the last commit is:
commit 2fe9cab33250d8e4e366d29c02c057814ff5523d
Author: Paul Eggert <egg...@cs.ucla.edu>
Date: Tue Jul 17 16:13:46 2018 -0700
It seems the getmntinfo() call is handled both for 4.4BSD and NetBSD
3.0, which is promising.
As a progress update (after I've implemented getmntinfo() on Threos):
findutils is no longer fails due to the mysterious struct fsstat (that
part still not works, but the configure could choose an alternate
implementation). It looks gnulib is happy with getmntinfo(), but I think
there is messup in the getfsstat() branch. I grepped on the entire
source code to find out where/how 'struct fsstat' is used, but I found
the "fsstat" (as a whole word) only in Changelog and m4 files. I would
guess the raw getfsstat() branch is for some really exotic system. For
now, I stop to figure out why struct fsstat is used in the m4 files.
Since gnulib is happy with 4.4BSD style getmntinfo(), so do I. Thanks to
Bruno for the getmntinfo() suggestion. And thank you all for the
valuable help to solve my configuration issue.
The next task for me is to try to replace forks with posix_spawn in
findutils.
Aron
