Add optional public header checking to GN build

Add optional public header checking to GN build

You can invoke this either using the "--check" argument to the gen command (for use on buildbots) or by running "gn check" (for developer on-demand use).

This adds support for "public" headers for a target, and an optional "--check" flag to the gen command that implements checkdeps-like include checking. Basically if you include a file that's declared in the build, it has to be in the public section of one of your dependents (or that dependent doesn't have a public section, which implies everything is public).

This roughly maps to Blaze's behavior around the public headers.

Moves modp_b64 build file to main tree.

Remove modp_b64 hack for older versions of VC missing stdint (this included basictypes which caused a header check failure).

I also added some base dependencies and some other minor build changes to solve some other issues identified by the check. The remaining one is that a file in base/metrics depends on ipc (!)

BUG=
R=scottmg@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=262216

[scottmg, 2014-04-05 00:40:06]

lgtm

https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
File tools/gn/c_include_iterator.cc (right):

https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
tools/gn/c_include_iterator.cc:40: // We on't handle C-style "//" comments since
this is the normal commenting
"only"

https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
tools/gn/c_include_iterator.cc:123: lines_since_last_include_++;
needs to get reset, or renamed. and a test for this skippy/early out behaviour i
guess.

https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check.cc
File tools/gn/command_check.cc (right):

https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check....
tools/gn/command_check.cc:29: if (!setup->DoSetup("//out/Default"))
//out/Check instead?

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker.cc
File tools/gn/header_checker.cc (right):

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
tools/gn/header_checker.cc:26: pending_files_(0) {
is this necessary? isn't there just a pool.Join that could be followed by Quit?

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
tools/gn/header_checker.cc:49: type != SOURCE_M && type != SOURCE_MM)
SOURCE_RC too, strictly speaking

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
tools/gn/header_checker.cc:174: // not unusual foe the buildfiles to not specify
that header at all.
foe->for

[brettw, 2014-04-07 05:08:19]

https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check.cc
File tools/gn/command_check.cc (right):

https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check....
tools/gn/command_check.cc:29: if (!setup->DoSetup("//out/Default"))
I'm currently using this for all the "temporary" generations like "args" and
"desc". We should change them in sync.

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker.cc
File tools/gn/header_checker.cc (right):

https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
tools/gn/header_checker.cc:26: pending_files_(0) {
Yes! It's actually the shutdown call I already make :-/  I guess I'll delete a
bunch of code.

[sa2213476, 2014-04-07 05:08:34]

On 2014/04/05 00:40:06, scottmg wrote:
> lgtm
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
> File tools/gn/c_include_iterator.cc (right):
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
> tools/gn/c_include_iterator.cc:40: // We on't handle C-style "//" comments
since
> this is the normal commenting
> "only"
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/c_include_iter...
> tools/gn/c_include_iterator.cc:123: lines_since_last_include_++;
> needs to get reset, or renamed. and a test for this skippy/early out behaviour
i
> guess.
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check.cc
> File tools/gn/command_check.cc (right):
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/command_check....
> tools/gn/command_check.cc:29: if (!setup->DoSetup("//out/Default"))
> //out/Check instead?
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker.cc
> File tools/gn/header_checker.cc (right):
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
> tools/gn/header_checker.cc:26: pending_files_(0) {
> is this necessary? isn't there just a pool.Join that could be followed by
Quit?
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
> tools/gn/header_checker.cc:49: type != SOURCE_M && type != SOURCE_MM)
> SOURCE_RC too, strictly speaking
> 
>
https://codereview.chromium.org/216903004/diff/120001/tools/gn/header_checker...
> tools/gn/header_checker.cc:174: // not unusual foe the buildfiles to not
specify
> that header at all.
> foe->for

[scottmg, 2014-04-07 16:19:50]

https://codereview.chromium.org/216903004/diff/140001/tools/gn/c_include_iter...
File tools/gn/c_include_iterator.cc (right):

https://codereview.chromium.org/216903004/diff/140001/tools/gn/c_include_iter...
tools/gn/c_include_iterator.cc:119: lines_since_last_include_ = 0;
LGTM with test for this.

[brettw, 2014-04-07 22:17:40]

Committed patchset #9 manually as r262216.