Message-ID: <1439490787.2114.36.camel@sipsolutions.net> (sfid-20150813_203316_706932_3EAB1269)
Subject: Re: [PATCH 0/2] iw: fixes to Android.mk to include "iw" in AOSP
 builds
From: Johannes Berg <johannes@sipsolutions.net>
To: Filipe Brandenburger <filbranden@google.com>
Cc: Arik Nemtsov <arik@wizery.com>, linux-wireless@vger.kernel.org,
	Elliott Hughes <enh@google.com>,
	Christopher Wiley <wiley@google.com>,
	Ying Wang <wangying@google.com>
Date: Thu, 13 Aug 2015 20:33:07 +0200
In-Reply-To: <CADU+-uDS2C4w-dcbR8Rxx+jRfG7kNvu8JTncH-fxDnSp5NOBfQ@mail.gmail.com> (sfid-20150813_185558_560372_445AA122)
References: <1438292115-39495-1-git-send-email-filbranden@google.com>
	 <1439456484.2114.6.camel@sipsolutions.net>
	 <CADU+-uDS2C4w-dcbR8Rxx+jRfG7kNvu8JTncH-fxDnSp5NOBfQ@mail.gmail.com>
	 (sfid-20150813_185558_560372_445AA122)
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Sender: linux-wireless-owner@vger.kernel.org

On Thu, 2015-08-13 at 09:55 -0700, Filipe Brandenburger wrote:

> Long term, considering this is in AOSP and is part of eng/userdebug
> builds, it's likely to be automatically available in future versions
> of Android.

Yeah, but most vendors don't actually build/run/ship AOSP nor
necessarily an unmodified iw, so it's likely not all that helpful.

> We ended up having to tweak the Android.mk again in the AOSP tree,
> since the current one is doing an "include" of the external Makefile,
> which in turns overrides the default "clean" rule and leaks some
> pattern rules such as %.o, etc. which affect the rest of Android
> build. For more details, please take a look at this Gerrit:
> https://android-review.googlesource.com/166104
> 
> I'm cc'ing Ying who spotted the problem and suggested the changes.

Thanks, but I'm not going to apply that patch in my tree - I don't want
to have to worry about the duplication of LOCAL_SRC_FILES.

> One problem with the current approach is that there's duplication of
> the names of the source files between Android.mk and Makefile, I was
> thinking perhaps we could introduce a common.mk or sources.mk that's
> included from both and only has the variables that can be used by
> both? Suggestions are welcome.

That seems reasonable.

> As mentioned in the initial thread, one option is to drop Android.mk
> from upstream altogether and maintain it in our AOSP downstream.

Given what I just said above about vendor builds etc. not being AOSP,
that probably wouldn't be all that helpful; I'm pretty sure it'd make
our life more difficult so I don't really like that idea much.

johannes