Add a small tool to answer questions like "Why does target A depend on target B".

Add a small tool to answer questions like "Why does target A depend on target B".

BUG=none
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=111430

[M-A Ruel, 2011-11-23 20:06:30]

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode2
tools/gyp-explain.py:2: 
Copyright

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode53
tools/gyp-explain.py:53: if __name__ == '__main__':
Use a main()

[Nico, 2011-11-23 20:12:07]

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode2
tools/gyp-explain.py:2: 
On 2011/11/23 20:06:30, Marc-Antoine Ruel wrote:
> Copyright

Done.

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode53
tools/gyp-explain.py:53: if __name__ == '__main__':
On 2011/11/23 20:06:30, Marc-Antoine Ruel wrote:
> Use a main()

Done.

[viettrungluu, 2011-11-23 20:19:51]

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode16
tools/gyp-explain.py:16: tools/gyp-explain.sh chrome_dll gtest#
Drive-by: ".sh"?

It might be more helpful to put:

Usage:
  ... <from_target> <to_target>

Example:
  ... chrome_dll gtest

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode19
tools/gyp-explain.py:19: def GetPath(graph, fro, to, get_all=False):
Is get_all even necessary, given that you're returning a generator?

[Nico, 2011-11-23 20:22:58]

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode16
tools/gyp-explain.py:16: tools/gyp-explain.sh chrome_dll gtest#
On 2011/11/23 20:19:51, viettrungluu wrote:
> Drive-by: ".sh"?

Done.

> It might be more helpful to put:
> 
> Usage:
>   ... <from_target> <to_target>
> 
> Example:
>   ... chrome_dll gtest

Not done.

http://codereview.chromium.org/8672006/diff/5001/tools/gyp-explain.py#newcode19
tools/gyp-explain.py:19: def GetPath(graph, fro, to, get_all=False):
On 2011/11/23 20:19:51, viettrungluu wrote:
> Is get_all even necessary, given that you're returning a generator? 

Oh, good point. Removed.

[M-A Ruel, 2011-11-23 20:28:04]

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode2
tools/gyp-explain.py:2: 
Most script don't put an empty line here, in fact I removed it everywhere since
it was the minority.

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode72
tools/gyp-explain.py:72: print '    GYP_GENERATORS=dump_dependency_json
build/gyp_chromium'
Why not run it automatically? It'd be more user friendly.

Note that when I run that, I get:
KeyError: 'Undefined variable RULE_INPUT_DIRNAME in chrome/chrome.gyp while
loading dependencies of chrome/browser/sync/tools/sync_tools.gyp while loading
dependencies of build/all.gyp while trying to load build/all.gyp'

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode74
tools/gyp-explain.py:74: sys.exit(1)
I highly prefer to return 1 and have
  sys.exit(Main(sys.argv))
as the last script line.

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode91
tools/gyp-explain.py:91: print 'No paths found from %s to %s.' % (fro, to)
return 1 so this can be used as a test?

[viettrungluu, 2011-11-23 20:30:40]

LGTM on my part, but maruel has more comments, apparently.

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py#newcode29
tools/gyp-explain.py:29: while len(q) > 0:
nit: |while q:|

[Nico, 2011-11-23 20:33:01]

Thanks, guys!

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode2
tools/gyp-explain.py:2: 
On 2011/11/23 20:28:05, Marc-Antoine Ruel wrote:
> Most script don't put an empty line here, in fact I removed it everywhere
since
> it was the minority.

Done.

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode72
tools/gyp-explain.py:72: print '    GYP_GENERATORS=dump_dependency_json
build/gyp_chromium'
On 2011/11/23 20:28:05, Marc-Antoine Ruel wrote:
> Why not run it automatically? It'd be more user friendly.

I disagree.

> Note that when I run that, I get:
> KeyError: 'Undefined variable RULE_INPUT_DIRNAME in chrome/chrome.gyp while
> loading dependencies of chrome/browser/sync/tools/sync_tools.gyp while loading
> dependencies of build/all.gyp while trying to load build/all.gyp'

Yeah, dump_dependency_json used to be broken everywhere except linux. I fixed
that yesterday, but I haven't rolled gyp to pick that up in chromium yet.

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode74
tools/gyp-explain.py:74: sys.exit(1)
On 2011/11/23 20:28:05, Marc-Antoine Ruel wrote:
> I highly prefer to return 1 and have
>   sys.exit(Main(sys.argv))
> as the last script line.

Yeah, me too, but I already sys.exit from MatchNode(). Doing It Right increases
the line count considerably, and all that check in clean up already almost
doubled its line count.

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode91
tools/gyp-explain.py:91: print 'No paths found from %s to %s.' % (fro, to)
On 2011/11/23 20:28:05, Marc-Antoine Ruel wrote:
> return 1 so this can be used as a test?

Can be added once something needs it :-)

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py#newcode29
tools/gyp-explain.py:29: while len(q) > 0:
On 2011/11/23 20:30:40, viettrungluu wrote:
> nit: |while q:|

Done.

[commit-bot: I haz the power, 2011-11-23 20:34:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/8672006/5003

[commit-bot: I haz the power, 2011-11-23 20:34:17]

Presubmit check for 8672006-5003 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit Warnings **
License must match:
.*? Copyright \(c\) 2011 The Chromium Authors\. All rights reserved\.\n.*? Use
of this source code is governed by a BSD-style license that can be\n.*? found in
the LICENSE file\.\n
Found a bad license header in these files:
  tools/gyp-explain.py

[M-A Ruel, 2011-11-23 20:35:57]

lgtm anyway

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode72
tools/gyp-explain.py:72: print '    GYP_GENERATORS=dump_dependency_json
build/gyp_chromium'
On 2011/11/23 20:33:01, Nico wrote:
> Yeah, dump_dependency_json used to be broken everywhere except linux. I fixed
> that yesterday, but I haven't rolled gyp to pick that up in chromium yet.

Ah, why?

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/5003/tools/gyp-explain.py#newcode51
tools/gyp-explain.py:51: sys.exit(1)
I didn't mean to remove the calls in this functions, just in Main(). Anyway, no
big deal.

[Nico, 2011-11-23 20:37:30]

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode72
tools/gyp-explain.py:72: print '    GYP_GENERATORS=dump_dependency_json
build/gyp_chromium'
On 2011/11/23 20:35:57, Marc-Antoine Ruel wrote:
> On 2011/11/23 20:33:01, Nico wrote:
> > Yeah, dump_dependency_json used to be broken everywhere except linux. I
fixed
> > that yesterday, but I haven't rolled gyp to pick that up in chromium yet.
> 
> Ah, why?

No particular reason. I'll roll it now.

[commit-bot: I haz the power, 2011-11-23 20:39:35]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/8672006/7004

[Mark Mentovai, 2011-11-23 21:01:23]

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py
File tools/gyp-explain.py (right):

http://codereview.chromium.org/8672006/diff/7001/tools/gyp-explain.py#newcode2
tools/gyp-explain.py:2: 
I actually prefer the empty line because it lets the boilerplate lice stand off
on its own.

Let’s bikeshed.

[commit-bot: I haz the power, 2011-11-23 22:25:53]

Try job failure for 8672006-7004 (retry) on mac_rel for step "media_unittests".
It's a second try, previously, step "media_unittests" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...