Re: [PATCHv2] add Android support

From: Jeff King <peff@peff.net>
Date: May 19, 2011, 12:20 UTC
Message-ID: <20110519122026.GA30240@sigill.intra.peff.net>
In-Reply-To: <2218C353-80FC-4540-A60E-608385384FB5@gieschke.de>

On Thu, May 19, 2011 at 01:37:53PM +0200, Rafael Gieschke wrote:

Show 12 quoted lines

> > (3) Add uname_S = Android (Makefile).
> > 
> > The first two would become much easier to justify if presented that
> > way. At least you won't hear from anybody "we don't want that much code to
> > not to run git on a phone!", as it is not entirely implausible to imagine
> > environments without support for one or both of these two facilities.
> 
> So, you would prefer to leave out ANDROID and use something like "ifeq
> ($(uname_S),Android)", so you will have to compile using make
> uname_S=Android? I would be fine with that, too. But I would also be
> fine with having to specify the build options on the command line or
> using a config.mak if you want to keep Android out of the Makefile.

The point of uname_S is that it would be found automatically. Sadly, There is nothing helpful in uname to tell us that we are on android:

  $ uname -a
  Linux localhost 2.6.37.4-cyanogenmod-01332-g7f230e8 #1 PREEMPT Tue Apr
  12 12:54:14 EDT 2011 armv7l GNU/Linux

You could obviously guess from Linux on that architecture, but that seems flaky to me. You can also figure it out by looking around the filesystem, but that is not something I'm excited about having the Makefile do.

So I think we are probably stuck either with the user setting an ANDROID meta-flag that sets the other flags appropriately, or leaving it up to the user to provide a sane config.mak.

-Peff