Call install-build-deps.sh from install-build-deps-android.sh

Call install-build-deps.sh from install-build-deps-android.sh

install-build-deps-android.sh relies on the assumption that
install-build-deps.sh has been previously invoked. At this point,
is makes sense for the Android script itself invoking its parent
dependency. 
See chromium-dev discussion: http://goo.gl/dKDX3R

BUG=
NOTRY=true

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=261535

[Torne, 2014-04-02 13:45:43]

This shouldn't be necessary, no? It's already in install-build-deps.sh and I
don't think it's expected that install-build-deps-android.sh is run *instead* of
that. As far as I know Android builds presuppose you are already set up to do
LInux builds, and the top of
https://code.google.com/p/chromium/wiki/AndroidBuildInstructions says so.

[himanshu.mistri, 2014-04-02 14:06:27]

On 2014/04/02 13:45:43, Torne wrote:
> This shouldn't be necessary, no? It's already in install-build-deps.sh and I
> don't think it's expected that install-build-deps-android.sh is run *instead*
of
> that. As far as I know Android builds presuppose you are already set up to do
> LInux builds, and the top of
> https://code.google.com/p/chromium/wiki/AndroidBuildInstructions says so.

Yes you may be right that it cover in install-build-deps.sh but if some one
follow only 

https://sites.google.com/a/chromium.org/dev/developers/how-tos/get-the-code

and then for android : fetch android --nosvn=True
so in that guide there may be happen that we will not run
./build/install-build-deps.sh.

[Primiano Tucci (use gerrit), 2014-04-02 14:09:25]

Right, Torne and I just had a discussion offline.
Uploaded a 2nd patchset to just invoke the Linux script from the Android one.

[Primiano Tucci (use gerrit), 2014-04-03 20:15:50]

Nico, would you mind taking a look to this 2-lines CL?
It is to strive to make the Android 1st build smoother.
Thanks

[Nico, 2014-04-03 20:19:56]

lgtm, makes sense, thanks!

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
File build/install-build-deps-android.sh (right):

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
build/install-build-deps-android.sh:21: "$(dirname
"${BASH_SOURCE[0]}")/install-build-deps.sh" \
Is `$(dirname "${0}") enough? It's not like there's several levels of sourcing
here.

[Nico, 2014-04-03 20:20:14]

(maybe reference the chromium-dev thread somewhere in the cl description)

[Primiano Tucci (use gerrit), 2014-04-03 20:54:10]

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
File build/install-build-deps-android.sh (right):

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
build/install-build-deps-android.sh:21: "$(dirname
"${BASH_SOURCE[0]}")/install-build-deps.sh" \
On 2014/04/03 20:19:56, Nico wrote:
> Is `$(dirname "${0}") enough? It's not like there's several levels of sourcing
> here.

I think the difference between $0 vs $BASH_SOURCE is that $BASH_SOURCE is safer,
in the sense that it should behave like __file__in python.
Essentially, I fear that if you run the script trhough ". install-build-deps.sh"
oppposite to "./install... or ./build/install...." that might fail.

[Nico, 2014-04-03 20:58:36]

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
File build/install-build-deps-android.sh (right):

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
build/install-build-deps-android.sh:21: "$(dirname
"${BASH_SOURCE[0]}")/install-build-deps.sh" \
On 2014/04/03 20:54:10, Primiano Tucci wrote:
> On 2014/04/03 20:19:56, Nico wrote:
> > Is `$(dirname "${0}") enough? It's not like there's several levels of
sourcing
> > here.
> 
> I think the difference between $0 vs $BASH_SOURCE is that $BASH_SOURCE is
safer,
> in the sense that it should behave like __file__in python.
> Essentially, I fear that if you run the script trhough ".
install-build-deps.sh"
> oppposite to "./install... or ./build/install...." that might fail.

We use $0 in a bunch of places, and I never heard of problems like this.

[Primiano Tucci (use gerrit), 2014-04-03 21:30:31]

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
File build/install-build-deps-android.sh (right):

https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
build/install-build-deps-android.sh:21: "$(dirname
"${BASH_SOURCE[0]}")/install-build-deps.sh" \
On 2014/04/03 20:58:36, Nico wrote:
> On 2014/04/03 20:54:10, Primiano Tucci wrote:
> > On 2014/04/03 20:19:56, Nico wrote:
> > > Is `$(dirname "${0}") enough? It's not like there's several levels of
> sourcing
> > > here.
> > 
> > I think the difference between $0 vs $BASH_SOURCE is that $BASH_SOURCE is
> safer,
> > in the sense that it should behave like __file__in python.
> > Essentially, I fear that if you run the script trhough ".
> install-build-deps.sh"
> > oppposite to "./install... or ./build/install...." that might fail.
> 
> We use $0 in a bunch of places, and I never heard of problems like this.

~ $ cat foo
#!/bin/bash
echo My dir-1 is $(dirname "${BASH_SOURCE[0]}")/
echo My dir-2 is $(dirname "$0")/

/tmp $ ./foo
My dir-1 is ./
My dir-2 is ./

/tmp $ . foo
My dir-1 is ./
My dir-2 is /bin/ #Ha wrong

/tmp $ /tmp/foo
My dir-1 is /tmp/
My dir-2 is /tmp/

/tmp $ . /tmp/foo
My dir-1 is /tmp/
My dir-2 is /bin/  # Ha wrong again!!!
/tmp $ 

That's it! I think just nobody complained, but BASH_SOURCE is IMHO actually more
reliable.
The real question is why we keep writing bash scripts, but that's another story
:)

[Primiano Tucci (use gerrit), 2014-04-03 21:34:35]

The CQ bit was checked by primiano@chromium.org

[commit-bot: I haz the power, 2014-04-03 21:36:33]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/primiano@chromium.org/222183002/20001

[commit-bot: I haz the power, 2014-04-03 21:40:20]

Change committed as 261535

[Nico, 2014-04-03 21:42:53]

On 2014/04/03 21:30:31, Primiano Tucci wrote:
>
https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
> File build/install-build-deps-android.sh (right):
> 
>
https://codereview.chromium.org/222183002/diff/20001/build/install-build-deps...
> build/install-build-deps-android.sh:21: "$(dirname
> "${BASH_SOURCE[0]}")/install-build-deps.sh" \
> On 2014/04/03 20:58:36, Nico wrote:
> > On 2014/04/03 20:54:10, Primiano Tucci wrote:
> > > On 2014/04/03 20:19:56, Nico wrote:
> > > > Is `$(dirname "${0}") enough? It's not like there's several levels of
> > sourcing
> > > > here.
> > > 
> > > I think the difference between $0 vs $BASH_SOURCE is that $BASH_SOURCE is
> > safer,
> > > in the sense that it should behave like __file__in python.
> > > Essentially, I fear that if you run the script trhough ".
> > install-build-deps.sh"
> > > oppposite to "./install... or ./build/install...." that might fail.
> > 
> > We use $0 in a bunch of places, and I never heard of problems like this.
> 
> ~ $ cat foo
> #!/bin/bash
> echo My dir-1 is $(dirname "${BASH_SOURCE[0]}")/
> echo My dir-2 is $(dirname "$0")/
> 
> /tmp $ ./foo
> My dir-1 is ./
> My dir-2 is ./
> 
> /tmp $ . foo
> My dir-1 is ./
> My dir-2 is /bin/ #Ha wrong
> 
> /tmp $ /tmp/foo
> My dir-1 is /tmp/
> My dir-2 is /tmp/
> 
> /tmp $ . /tmp/foo
> My dir-1 is /tmp/
> My dir-2 is /bin/  # Ha wrong again!!!
> /tmp $ 
> 
> That's it! I think just nobody complained, but BASH_SOURCE is IMHO actually
more
> reliable.

No, that's exactly my point: This script isn't intended to be sourced.

> The real question is why we keep writing bash scripts, but that's another
story
> :)

Agreed, but the Real Evil are bash scripts that have to be sourced instead of
run (runnable bash scripts can be replaced with python once they get longer).
BASH_SOURCE is only useful for sourceable scripts from what I understand.