Re: Intent to ship: Array.prototype.values

[Tom Schuster]

I plan on landing this today, so this change is going to be in Firefox
60. The pref for enabling/disabling this feature is
javascript.options.array_prototype_values.

On Fri, Feb 2, 2018 at 4:20 PM, Tom Schuster  wrote:
> I talked to   Jan de Mooij and it's feasible to add a pref for this,
> so we are definitely going to do this.
>
> Array.prototype.values has always been enabled on Nightly. Based on
> where the breakage occurred, internal company webpages, I would not
> expect EARLY_BETA_OR_EARLIER to shake out any additional problems.
>
> On Fri, Feb 2, 2018 at 4:17 PM, Mike Taylor  wrote:
>> On Feb 2, 2018, at 7:45 AM, Tom Schuster  wrote:
>>> Any additional ideas how to mitigate the risk here? Chrome seems to
>>> want to add a kill pref for this,  from my experience more difficult
>>> for us with the way we define methods in SpiderMonkey. Should that be
>>> a requirement for shipping?
>>
>> Add a pref enabling it for EARLY_BETA_OR_EARLIER, and false otherwise, and 
>> let’s see what Nightly/Beta shakes out for a few releases. Then we have time 
>> to attempt outreach or other such hacks before burning our release users.
>>
>> --
>> Mike Taylor
>> Web Compat, Mozilla
>>
>>
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Emilio Cobos Álvarez]

On 02/02/2018 03:27 PM, Andrew Overholt wrote:
> On Fri, Feb 2, 2018 at 8:45 AM, Tom Schuster  > wrote:
> 
> Chrome seems to
> want to add a kill pref for this,  from my experience more difficult
> for us with the way we define methods in SpiderMonkey. Should that be
> a requirement for shipping?
> 
> 
> If it's not too difficult it does make it a lot easier to fix things if
> it causes major breakage. I think we can even do a pref flip without
> shipping an update which is pretty nice.
> 
> Is there a way to do per-site pref changes? Did we do this for Stylo,
> Emilio?

We did, but it required a fair amount of machinery. The code for the
stylo domain blocklist was removed very recently in bug 1426223.

 -- Emilio
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Tom Schuster]

I talked to   Jan de Mooij and it's feasible to add a pref for this,
so we are definitely going to do this.

Array.prototype.values has always been enabled on Nightly. Based on
where the breakage occurred, internal company webpages, I would not
expect EARLY_BETA_OR_EARLIER to shake out any additional problems.

On Fri, Feb 2, 2018 at 4:17 PM, Mike Taylor  wrote:
> On Feb 2, 2018, at 7:45 AM, Tom Schuster  wrote:
>> Any additional ideas how to mitigate the risk here? Chrome seems to
>> want to add a kill pref for this,  from my experience more difficult
>> for us with the way we define methods in SpiderMonkey. Should that be
>> a requirement for shipping?
>
> Add a pref enabling it for EARLY_BETA_OR_EARLIER, and false otherwise, and 
> let’s see what Nightly/Beta shakes out for a few releases. Then we have time 
> to attempt outreach or other such hacks before burning our release users.
>
> --
> Mike Taylor
> Web Compat, Mozilla
>
>
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Mike Taylor]

On Feb 2, 2018, at 7:45 AM, Tom Schuster  wrote:
> Any additional ideas how to mitigate the risk here? Chrome seems to
> want to add a kill pref for this,  from my experience more difficult
> for us with the way we define methods in SpiderMonkey. Should that be
> a requirement for shipping?

Add a pref enabling it for EARLY_BETA_OR_EARLIER, and false otherwise, and 
let’s see what Nightly/Beta shakes out for a few releases. Then we have time to 
attempt outreach or other such hacks before burning our release users.

--
Mike Taylor
Web Compat, Mozilla


___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Lars Hansen]

On Fri, Feb 2, 2018 at 2:45 PM, Tom Schuster  wrote:

> Any additional ideas how to mitigate the risk here? Chrome seems to
> want to add a kill pref for this,  from my experience more difficult
> for us with the way we define methods in SpiderMonkey. Should that be
> a requirement for shipping?
>

​Don't we need that type of availability mechanism to support
HTTPS-everywhere anyway?  "All new features..."

--lars
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Boris Zbarsky]

On 2/2/18 8:45 AM, Tom Schuster wrote:

Should that be a requirement for shipping?


Imo, yes, given the history of trying to ship this in the past (had to 
be backed out for web compat fail twice, etc).


-Boris
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Re: Intent to ship: Array.prototype.values

[Andrew Overholt]

On Fri, Feb 2, 2018 at 8:45 AM, Tom Schuster  wrote:

> Chrome seems to
> want to add a kill pref for this,  from my experience more difficult
> for us with the way we define methods in SpiderMonkey. Should that be
> a requirement for shipping?
>

If it's not too difficult it does make it a lot easier to fix things if it
causes major breakage. I think we can even do a pref flip without shipping
an update which is pretty nice.

Is there a way to do per-site pref changes? Did we do this for Stylo,
Emilio?
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform

Intent to ship: Array.prototype.values

[Tom Schuster]

We already tried to ship Array.prototype.values before in Firefox 48,
but this broke
Microsoft Dynamics CRM 2011
(https://bugzilla.mozilla.org/show_bug.cgi?id=1299593). We are however
aware this might have caused other breakages as well.

The compatibility risk for this change is high.
However Chrome is going to try shipping this again (third time!)
https://groups.google.com/a/chromium.org/forum/#!topic/blink-dev/yX4U__z67xo

Edge and Safari already ship this feature.

Bug to turn on by default: https://bugzilla.mozilla.org/show_bug.cgi?id=1420101

Any additional ideas how to mitigate the risk here? Chrome seems to
want to add a kill pref for this,  from my experience more difficult
for us with the way we define methods in SpiderMonkey. Should that be
a requirement for shipping?
___
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform