Re: [webkit-dev] Experimental features enabled by default

2018-06-27 Thread Michael Catanzaro
On Wed, Jun 27, 2018 at 1:00 PM, Tim Horton  
wrote:

Alright, how’s it looking now? :)


Good, thanks for taking care of this, Tim!

Michael

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


Re: [webkit-dev] Experimental features enabled by default

2018-06-27 Thread Tim Horton
Alright, how’s it looking now? :)

> On Jun 26, 2018, at 19:06, Tim Horton  wrote:
> 
> 
> 
>> On Jun 26, 2018, at 18:55, Michael Catanzaro  wrote:
>> 
>> 
>> Hi,
>> 
>> We're a lot closer after r233227!
>> 
>> There is still IsSecureContextAttributeEnabled. Shall we switch the default 
>> value of that one to use DEFAULT_EXPERIMENTAL_FEATURES_ENABLED instead of 
>> true?
> 
> Not sure, waiting for other people to get back to me. There’s some indication 
> that this one, indeed, should follow DEFAULT_EXPERIMENTAL_FEATURES_ENABLED, 
> but I want to make sure :)
> 
>> We should also move CSSAnimationTriggersEnabled to the experimental features 
>> section of the file, and add the category tag. FullScreenEnabled and 
>> EncryptedMediaAPIEnabled should also be moved down in the file.
> 
> Will have to peek at these!
> 
>> Then we can just update the comment to indicate that custom values like 
>> DEFAULT_SERVICE_WORKERS_ENABLED are fine as long as they're thoughtful and 
>> consider ENABLE(EXPERIMENTAL_FEATURES), and that will take care of the other 
>> cases.
> 
> Makes sense.
> 
>> Michael
>> 
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


Re: [webkit-dev] Experimental features enabled by default

2018-06-26 Thread Tim Horton


> On Jun 26, 2018, at 18:55, Michael Catanzaro  wrote:
> 
> 
> Hi,
> 
> We're a lot closer after r233227!
> 
> There is still IsSecureContextAttributeEnabled. Shall we switch the default 
> value of that one to use DEFAULT_EXPERIMENTAL_FEATURES_ENABLED instead of 
> true?

Not sure, waiting for other people to get back to me. There’s some indication 
that this one, indeed, should follow DEFAULT_EXPERIMENTAL_FEATURES_ENABLED, but 
I want to make sure :)

> We should also move CSSAnimationTriggersEnabled to the experimental features 
> section of the file, and add the category tag. FullScreenEnabled and 
> EncryptedMediaAPIEnabled should also be moved down in the file.

Will have to peek at these!

> Then we can just update the comment to indicate that custom values like 
> DEFAULT_SERVICE_WORKERS_ENABLED are fine as long as they're thoughtful and 
> consider ENABLE(EXPERIMENTAL_FEATURES), and that will take care of the other 
> cases.

Makes sense.

> Michael
> 
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


Re: [webkit-dev] Experimental features enabled by default

2018-06-26 Thread Michael Catanzaro



Hi,

We're a lot closer after r233227!

There is still IsSecureContextAttributeEnabled. Shall we switch the 
default value of that one to use DEFAULT_EXPERIMENTAL_FEATURES_ENABLED 
instead of true?


We should also move CSSAnimationTriggersEnabled to the experimental 
features section of the file, and add the category tag. 
FullScreenEnabled and EncryptedMediaAPIEnabled should also be moved 
down in the file.


Then we can just update the comment to indicate that custom values like 
DEFAULT_SERVICE_WORKERS_ENABLED are fine as long as they're thoughtful 
and consider ENABLE(EXPERIMENTAL_FEATURES), and that will take care of 
the other cases.


Michael

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


Re: [webkit-dev] Experimental features enabled by default

2018-06-20 Thread Michael Catanzaro

On Wed, Jun 20, 2018 at 6:05 PM, Tim Horton  wrote:

It’s definitely not OK, no.


Well, we should make sure the right choice is made for each individual 
feature, but I don't know what those right choices are... hence the 
motivation for my mail.


I think we have some sorting out of this to do, I think it would be 
appreciated if you could hold off for a little bit. Sorry for the 
mess!


OK, I didn't know you were already discussing this. I'll hold off, then.

Michael

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


Re: [webkit-dev] Experimental features enabled by default

2018-06-20 Thread Tim Horton
Resending from the right address.

> On Jun 20, 2018, at 11:12 AM, Michael Catanzaro  wrote:
> 
> Hi,

Apple folks have been chatting about this elsewhere, so it’s a humorous 
coincidence to see this here at the same time.

> I noticed in glancing over WebPreferences.yaml that we have a large number of 
> experimental features enabled by default. There is a comment header 
> indicating that the only allowed values are false or 
> DEFAULT_EXPERIMENTAL_FEATURES_ENABLED, so true is disallowed, but that's 
> being ignored in many cases. It's become a bit difficult to notice the 
> comment header because the experimental feature list has grown quite large in 
> the past year or two.
> 
> Background: DEFAULT_EXPERIMENTAL_FEATURES_ENABLED is enabled by default in 
> the XCode build (because Apple disables experimental features on branches, 
> not trunk), disabled by default in the CMake build (to prevent experimental 
> features from sneaking into GTK and WPE releases), and finally enabled by 
> build-webkit (so experimental features are always enabled for bots and 
> developers).
> 
> I'm planning to change the default values of the following features from true 
> (not allowed for experimental features) to 
> DEFAULT_EXPERIMENTAL_FEATURES_ENABLED:
> 
> CacheAPIEnabled
> ConstantPropertiesEnabled
> CrossOriginWindowPolicySupportEnabled
> IsSecureContextAttributeEnabled
> StorageAccessAPIEnabled
> SubresourceIntegrityEnabled
> RestrictedHTTPResponseAccess
> CrossOriginResourcePolicyEnabled
> StorageAccessPromptsEnabled
> 
> But I wonder if this is OK for all of the above features, as no doubt the 
> intention behind using true was to make the feature always enabled.

It’s definitely not OK, no.

I think we have some sorting out of this to do, I think it would be appreciated 
if you could hold off for a little bit. Sorry for the mess!

> So if any of the above features are no longer experimental and should be 
> enabled by default on all ports, please speak up so they can be graduated out 
> of the experimental features category. My recommended sanity-check is that if 
> the feature is ready for all Safari users and not just Tech Preview users, 
> then it's probably no longer experimental.
> 
> A couple more oddities:
> 
> I'll mark CSSAnimationTriggersEnabled as experimental, since it uses 
> DEFAULT_EXPERIMENTAL_FEATURES_ENABLED and is clearly intended to be.
> 
> For DisabledAdaptationsMetaTagEnabled, I'm not sure. We could change the 
> default value from DISABLED_ADAPTATIONS_META_TAG_ENABLED to 
> DEFAULT_EXPERIMENTAL_FEATURES_ENABLED, and add a PLATFORM(WATCHOS) condition. 
> I think this should be OK for watchOS. Alternatively, I wonder if it might be 
> better to consider it non-experimental and simply set the default to true 
> (with a PLATFORM(WATCHOS) condition)?
> 
> Michael
> 
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev