Intent to unship: Request.context

[Ehsan Akhgari]

Tracking bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1188062

APIs to be removed: Request.context

This was shipped in Firefox 39, but this attribute is fairly useless
without service workers since it can only be read from Request objects and
its value will always be "fetch" for manually created Request objects.

It has become clear that the meaning of this attribute needs to change in
the fetch spec, so I think it is best for us to unship this attribute for
now, and expose it again once the spec has settled.

The web compat effect should be minimal if existent, as we have never
exposed any useful values from this attribute, and no other UAs ship it.

I'm planning to unship this in Firefox 42.

Cheers,
--
Ehsan

[Ben Kelly]

On Mon, Jul 27, 2015 at 5:55 PM, Ehsan Akhgari <ehsan....@gmail.com>
wrote:

> This was shipped in Firefox 39, but this attribute is fairly useless
> without service workers since it can only be read from Request objects and
> its value will always be "fetch" for manually created Request objects.

Actually, it sounds like we're not quite right there either. It should now
be "" for synthetically created Request objects. The fetch() method copies
to a new Request object with "fetch" context. Thats not observable without
the fetch event, though.

Anyway, that's beside the point right now if we are disabling it.

Ben

[fal...@chromium.org]

On Tuesday, July 28, 2015 at 6:56:40 AM UTC+9, Ehsan Akhgari wrote:
> The web compat effect should be minimal if existent, as we have never
> exposed any useful values from this attribute, and no other UAs ship it.

Just for reference, Chrome 44 shipped this too:
https://www.chromestatus.com/features/4699713102151680

It's probably still early enough to unship or change. I've updated the dashboard to point to this thread.

[Ehsan Akhgari]

On 2015-07-27 8:43 PM, fal...@chromium.org wrote:
> On Tuesday, July 28, 2015 at 6:56:40 AM UTC+9, Ehsan Akhgari wrote:
>> The web compat effect should be minimal if existent, as we have never
>> exposed any useful values from this attribute, and no other UAs ship it.
>
> Just for reference, Chrome 44 shipped this too:
> https://www.chromestatus.com/features/4699713102151680

Oh, thanks for the correction!

> It's probably still early enough to unship or change. I've updated the dashboard to point to this thread.

I think it's probably a good idea for Blink to also unship for now, or
at least plan to change the implementation according to the spec changes
(which most likely will be backwards incompatible.)

Note that the main reason why we decided to unship this was that if a
Request object is saved in the DOM Cache, it will almost definitely be
impossible to reflect the correct information on it when you read it out
of the cache again without knowing what the spec decides to do exactly
and save the necessary information for reflecting that inside the DOM Cache.