Fix IDL dependency computation for partial interfaces

Fix IDL dependency computation for partial interfaces

Current IDL dependencies consider partial interfaces as dependencies
for *all* IDL files (to avoid having to resolve it to a particular file).

The code has 2 mistakes, however:
1. It fails to do this for supplemental IDL files (notably testing),
  which have just gotten a partial interface.
2. There's a typo in a regex, which means it fails on simple IDL files.

This fixes both bugs, and simplifies the partial interface determination code.

This fixes the problem that required this revert:
Revert r160289 "Vibration cannot be canceled during pattern vibration."
https://codereview.chromium.org/30063003/

This should allow this CL to be relanded:
Vibration cannot be canceled during pattern vibration.
https://codereview.chromium.org/18478003/

To verify that this CL works (excluding testing),
verify that NavigationVibration.idl is a global dependency:
cd "$CHROMIUM_DIR"
regyp  # eg gclient runhooks
ninja -C out/Release -j100 all_webkit
touch third_party/WebKit/Source/modules/vibration/NavigatorVibration.idl
ninja -C out/Release -j100 all_webkit  # should rebuild all bindings

To verify that it has fixed the problem,
apply the patch and verify InternalVibration.idl is a global dependency:
cd "$BLINK_DIR"
git co -b reland-vibe
git cl patch 18478003

cd "$CHROMIUM_DIR"
regyp  # eg gclient runhooks
ninja -C out/Release -j100 all_webkit
touch third_party/WebKit/Source/modules/vibration/testing/InternalsVibration.idl
ninja -C out/Release -j100 all_webkit  # should rebuild all bindings

BUG=222504, 261467, 310137
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=160362

[Nils Barth (inactive), 2013-10-23 07:27:18]

Chris or haraken,
could y'all take a look at this please?
(Fixes Kihong's problem and improves dependency resolution some.)

[kihong, 2013-10-23 07:40:34]

https://codereview.chromium.org/29323008/diff/1/Source/bindings/derived_sourc...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/29323008/diff/1/Source/bindings/derived_sourc...
Source/bindings/derived_sources.gyp:204: '<@(main_idl_files)',
Don't we need to add support_idl_files here also?

[Nils Barth (inactive), 2013-10-23 08:13:17]

https://codereview.chromium.org/29323008/diff/1/Source/bindings/derived_sourc...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/29323008/diff/1/Source/bindings/derived_sourc...
Source/bindings/derived_sources.gyp:204: '<@(main_idl_files)',
On 2013/10/23 07:40:34, kihong wrote:
> Don't we need to add support_idl_files here also?

testing_support_idl_files includes support_idl_files,
so this is correct.
However, you're right that this is confusing;
I've rewritten it to make the sources/inputs & used variables
more transparent -- thanks!

[kihong, 2013-10-23 08:18:43]

https://codereview.chromium.org/29323008/diff/80001/Source/bindings/derived_s...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/29323008/diff/80001/Source/bindings/derived_s...
Source/bindings/derived_sources.gyp:267: '<@(support_idl_files)
<@(generated_support_idl_files)',
IMHO, we don't need support_idl_files here also?

[Nils Barth (inactive), 2013-10-23 10:29:00]

Thanks for comment Kihong, this clarifies and simplifies it!

https://codereview.chromium.org/29323008/diff/80001/Source/bindings/derived_s...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/29323008/diff/80001/Source/bindings/derived_s...
Source/bindings/derived_sources.gyp:267: '<@(support_idl_files)
<@(generated_support_idl_files)',
On 2013/10/23 08:18:44, kihong wrote:
> IMHO, we don't need support_idl_files here also?

Got it, removed, and updated comments to reflect these.

OIC:
Now that support_idl_files are included in InterfaceDependencies,
we can remove them from this list of additional files.
OTOH, we have to list the generated files, since they're not present
at GYP run time, so we don't include them in InterfaceDependencies.

We could include generated files in InterfaceDependencies,
and then we could get rid of additionalIdlFiles,
but this would require some hacking.

[haraken, 2013-10-23 12:15:39]

LGTM, but please wait for an approval from kihong.

[kihong, 2013-10-23 14:26:02]

lgtm

Thanks Nils.

[Michael van Ouwerkerk, 2013-10-23 17:02:59]

On 2013/10/23 14:26:02, kihong wrote:
> lgtm
> 
> Thanks Nils.

Thanks for the quick fix! Is this ready to submit then?

[Inactive, 2013-10-23 18:24:06]

lgtm

[commit-bot: I haz the power, 2013-10-23 18:24:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/29323008/210001

[commit-bot: I haz the power, 2013-10-23 19:21:44]

Change committed as 160362