Add a flag to generate the Makefile in a different directory than...

Add a flag to generate the Makefile in a different directory than
--depth.  This allows us to properly generate a Makefile in an
upstream checkout where we want to have the Makefile in the root
checkout, but we want <(DEPTH) to be at WebKit/chromium because
that's where gclient pulls our dependencies.

TEST=test/toplevel-dir/gyptest-toplevel-dir.py

[tony, 2010-05-14 13:10:05]

This is to try to fix the situation in the upstream build where the Makefile
isn't currently generated at the root of the tree.  makefile-dir specifies where
we want the Makefile to live (in our case, it'll be --makefile-dir=../../).

[Evan Martin, 2010-05-17 17:18:18]

Did you leave out a file?  Other then tests, I only see __init__.py change.

[tony, 2010-05-18 00:01:13]

On 2010/05/17 17:18:18, Evan Martin wrote:
> Did you leave out a file?  Other then tests, I only see __init__.py change.

Sorry, gcl user error on my part.  New patchset with changes to make.py
uploaded.

[Evan Martin, 2010-05-18 09:05:57]

This code looks good to me, though I am nervous it won't work in some subtle
way, like something assuming that depth is the root of the tree.

http://codereview.chromium.org/2122004/diff/6001/7001
File pylib/gyp/__init__.py (right):

http://codereview.chromium.org/2122004/diff/6001/7001#newcode270
pylib/gyp/__init__.py:270: help='where to put Makefile')
I wonder if this flag ought to have a more generic purpose, like "where to put
project files".

[tony, 2010-05-18 09:20:38]

On 2010/05/18 09:05:57, Evan Martin wrote:
> This code looks good to me, though I am nervous it won't work in some subtle
> way, like something assuming that depth is the root of the tree.

Probably.  It works with the current webkit build, so it seems ok for now.

> http://codereview.chromium.org/2122004/diff/6001/7001
> File pylib/gyp/__init__.py (right):
> 
> http://codereview.chromium.org/2122004/diff/6001/7001#newcode270
> pylib/gyp/__init__.py:270: help='where to put Makefile')
> I wonder if this flag ought to have a more generic purpose, like "where to put
> project files".

Yeah, I didn't like having a makefile specific flag exposed to gyp, but VS and
xcode don't have top level project files (well, VS does, but we generate one for
every gyp file which causes it to have lots of different output dirs).  I guess
scons has a top level project file, but does scons even work now?  Maybe
something like "toplevel-file-dir"?

[Evan Martin, 2010-05-18 09:25:02]

I wouldn't worry about scons.  LGTM+=

[Michael Moss, 2010-05-18 17:35:29]

> something like "toplevel-file-dir"?

Yeah, I like the generic name better (even in make, it could be useful for
something besides the Makefile).

[tony, 2010-05-19 01:02:58]

On 2010/05/18 17:35:29, Michael Moss wrote:
> > something like "toplevel-file-dir"?
> 
> Yeah, I like the generic name better (even in make, it could be useful for
> something besides the Makefile).

Ok, I've renamed the flag to toplevel-dir.  I also changed the debug output to
sort the options so it's more stable and consistent across python
implementations.

[tony, 2010-05-20 03:25:19]

Michael, if this looks good to you, can you commit it for me?