On Sun, 2018-06-17 at 22:31 +0200, Bruno Haible wrote:
> > I thought in general we were avoiding use of the _WINxx
> > preprocessor defines, and instead preferring WINDOWS32.
>
> Maybe in general in GNU. But not in gnulib.
>
> > > But there is no implementation for native Windows!
> >
> > True,
Hi Paul,
> probably not great QOI but that's what it does.
> ...
> Thanks for your link to the Microsoft docs, but I'm personally not
> going to undertake this effort. Maybe someone else will.
OK.
> > The proper condition for testing for a native Windows platform is
> > #if defined _WIN32 &&
Paul Smith wrote:
I could instead change code inside GNU make to not invoke getloadavg()
at all on Windows, but that seems less clean...
Your patch to the getloadavg module looks safe and helps GNU 'make' build. I
installed it since I'm the one who got rid of the HAVE_UNISTD_H #if long ago
(c
On Mon, 2018-05-21 at 01:26 +0200, Bruno Haible wrote:
> Hi Paul,
> The proper condition for testing for a native Windows platform is
> #if defined _WIN32 && ! defined __CYGWIN__
Does this mean you'd prefer to see a patch that tested this before
including unistd.h, versus the HAVE_UNISTD_H test
Hi Paul,
But there is no implementation for native Windows!
Cygwin does it by use of the PDH functions [1], plus some custom
logic to get the averages right. But 'make' needs only the first number,
therefore it would be sufficient to provide the _current_ load instead
of an average, right? In thi
* lib/getloadavg.c: Reinstate ifdef for HAVE_UNISTD_H
* m4/getloadavg.m4: Check for unistd.h
---
An alternative would be to use ifdef WINDOWS32 instead of checking
for unistd.h.
lib/getloadavg.c | 4 +++-
m4/getloadavg.m4 | 2 +-
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/lib/
