Add [NoImplHeader] extended attribute

Add [NoImplHeader] extended attribute

If all of attributes and methods of an interface are implemented in private scripts,
there is no header file in the C++ side. To prevent the IDL compiler from including
non-existing header files, this CL adds an IDL extended attribute to specify the fact.

BUG=341031
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=181093

[haraken, 2014-08-29 11:29:07]

haraken@chromium.org changed reviewers:
+ jl@opera.com, vivekg@chromium.org

[haraken, 2014-08-29 11:29:07]

PTAL

Another approach is to check the existence of the header file to know whether
the header file exists or not, but I'm not really a fan of the approach because
it might cause flaky behavior depending on actual contents of directories on
each environment (build bots etc).

[Jens Widell, 2014-08-29 12:17:04]

LGTM

On 2014/08/29 11:29:07, haraken wrote:
> Another approach is to check the existence of the header file to know whether
> the header file exists or not, but I'm not really a fan of the approach
because
> it might cause flaky behavior depending on actual contents of directories on
> each environment (build bots etc).

A third approach would be to check whether all attributes and methods in the
interface are implemented in private scripts. IOW, calculate whether we'll be
using anything from the header, and skip the include if not. (And not really
care whether the header exists.)

But like your alternative approach, this would be more work, and more
error-prone, and not add much value (at least not at this point.)

[haraken, 2014-08-29 12:19:40]

Thanks for review!

> A third approach would be to check whether all attributes and methods in the
> interface are implemented in private scripts. IOW, calculate whether we'll be
> using anything from the header, and skip the include if not. (And not really
> care whether the header exists.)
> 
> But like your alternative approach, this would be more work, and more
> error-prone, and not add much value (at least not at this point.)

Yeah, at first I tried the approach with tasak@ but gave up due to the
complexity...

[haraken, 2014-08-29 12:19:44]

The CQ bit was checked by haraken@chromium.org

[commit-bot: I haz the power, 2014-08-29 12:20:45]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/haraken@chromium.org/519703002/1

[commit-bot: I haz the power, 2014-08-29 13:25:38]

Committed patchset #1 (id:1) as 181093

[haraken, 2014-09-01 01:35:16]

haraken@chromium.org changed reviewers:
+ yosin@chromium.org

[haraken, 2014-09-01 01:35:17]

yosin, vivekg: You can use [NoImplHeader] and avoid adding an empty header file
in the Blink side.

[yosin_UTC9, 2014-09-01 03:37:46]

On 2014/09/01 01:35:17, haraken wrote:
> yosin, vivekg: You can use [NoImplHeader] and avoid adding an empty header
file
> in the Blink side.

Thanks!