Relax feature defines restrictions.

Relax feature defines restrictions.

Checks in fremontcutbuilder don't take into account nested conditions (e.g. PagePopupController.formatMonth).
The check in databasebuilder is too aggressive as we never use conditionals in dart.idl.

R=blois@google.com

Committed: https://code.google.com/p/dart/source/detail?r=22498

[blois, 2013-05-07 18:20:57]

lgtm

[podivilov, 2013-05-08 08:05:22]

Committed patchset #1 manually as r22498 (presubmit successful).

[Anton Muhin, 2013-05-13 09:28:01]

DBQ: Why this change?

People tend to ignore those warnings---that's exactly why I made the code throw.

[podivilov, 2013-05-13 10:15:33]

On 2013/05/13 09:28:01, Anton Muhin wrote:
> DBQ: Why this change?
> 
> People tend to ignore those warnings---that's exactly why I made the code
throw.

These checks might be incorrect in some circumstances (e.g. clank build where
more features are disabled), see issue description.

[Anton Muhin, 2013-05-13 10:22:49]

On 2013/05/13 10:15:33, podivilov wrote:
> On 2013/05/13 09:28:01, Anton Muhin wrote:
> > DBQ: Why this change?
> > 
> > People tend to ignore those warnings---that's exactly why I made the code
> throw.
> 
> These checks might be incorrect in some circumstances (e.g. clank build where
> more features are disabled), see issue description.

If it's a matter with Clank, I'd prefer to have Clank specific set of
conditionals instead, otherwise we're guaranteed to generate libs against
different set of conditionals.

[podivilov, 2013-05-13 11:51:07]

On 2013/05/13 10:22:49, Anton Muhin wrote:
> On 2013/05/13 10:15:33, podivilov wrote:
> > On 2013/05/13 09:28:01, Anton Muhin wrote:
> > > DBQ: Why this change?
> > > 
> > > People tend to ignore those warnings---that's exactly why I made the code
> > throw.
> > 
> > These checks might be incorrect in some circumstances (e.g. clank build
where
> > more features are disabled), see issue description.
> 
> If it's a matter with Clank, I'd prefer to have Clank specific set of
> conditionals instead, otherwise we're guaranteed to generate libs against
> different set of conditionals.

I'm not sure I understand what you mean. Clank does have a specific set of
conditionals. The actual problem is that these checks are wrong .

[Anton Muhin, 2013-05-13 11:53:38]

Why those checks are wrong?

On 2013/05/13 11:51:07, podivilov wrote:
> On 2013/05/13 10:22:49, Anton Muhin wrote:
> > On 2013/05/13 10:15:33, podivilov wrote:
> > > On 2013/05/13 09:28:01, Anton Muhin wrote:
> > > > DBQ: Why this change?
> > > > 
> > > > People tend to ignore those warnings---that's exactly why I made the
code
> > > throw.
> > > 
> > > These checks might be incorrect in some circumstances (e.g. clank build
> where
> > > more features are disabled), see issue description.
> > 
> > If it's a matter with Clank, I'd prefer to have Clank specific set of
> > conditionals instead, otherwise we're guaranteed to generate libs against
> > different set of conditionals.
> 
> I'm not sure I understand what you mean. Clank does have a specific set of
> conditionals. The actual problem is that these checks are wrong .