778044 - Add a way to pref off interface objects

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Boris Zbarsky [:bzbarsky]]

We need this for things like WebSocket.

We have two options here.  We can either make this call nsWebSocket::PrefEnabled(), and generally use that pattern, or put the actual pref in the IDL.

I'm leaning towards the former to not duplicate the pref name and whatnot.

Comment 1

[(no longer active)]

Putting the actual pref in the IDL would make things more readable, IMO.

Comment 2

[Boris Zbarsky [:bzbarsky]]

How so?  The IDL would be quite readable with:

  [PrefControlled]
  interface Foo {
  };

which would call FooNativeType::PrefEnabled() on the C++ side.

Comment 3

[(no longer active)]

I was talking about the name of the pref...  Of course I may not quite know what this bug was filed for... :-)

Comment 4

[Boris Zbarsky [:bzbarsky]]

Well, the name of the pref in my proposed setup would be in just one place: the PrefEnabled() impl.

Note that we have such an impl already and would still have it, so putting the pref name in the IDL would mean having it in two places...

Comment 5

[(no longer active)]

(In reply to comment #4)
> Well, the name of the pref in my proposed setup would be in just one place: the
> PrefEnabled() impl.
> 
> Note that we have such an impl already and would still have it, so putting the
> pref name in the IDL would mean having it in two places...

OK, what you're saying sounds good to me!

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

Could we autogenerate the pref check and have pref mentioned only in the webidl?

Comment 7

[Boris Zbarsky [:bzbarsky]]

At least in the websocket case the implementation methods actually check the pref in various places.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Peter, would it make any sense to just store a bool* in the namestruct and if non-null to read what it's pointing to?  Then we could initially point it to some boolean that observes the pref (or null, if not preffed).

Comment 9

[Boris Zbarsky [:bzbarsky]]

Attachment: Add a way to pref off Paris binding constructor objects.

Comment 10

[Boris Zbarsky [:bzbarsky]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/8d5589b88c8b

Comment 11

[Boris Zbarsky [:bzbarsky]]

Documented: https://developer.mozilla.org/en-US/docs/Mozilla/WebIDL_bindings#PrefControlled

Comment 12

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/8d5589b88c8b

Comment 13

[(no longer active)]

Is PrefEnabled called every time the value is looked up?

Comment 14

[Boris Zbarsky [:bzbarsky]]

If it keeps returning false, then yes.

If it returns true for a given window, it will normally not get called after that for that window.

Comment 15

[(no longer active)]

(In reply to comment #14)
> If it keeps returning false, then yes.
> 
> If it returns true for a given window, it will normally not get called after
> that for that window.

That's not great, at least for the purpose of writing tests for a preffed-off feature.  Can we fix that easily?

Comment 16

[Boris Zbarsky [:bzbarsky]]

Not without completely rewriting how prototypes are looked up on windows, unfortunately.

That said, the only problem arises if you have a test that takes a preffed-off feature, flips the pref, tests the feature, flips the pref back, and then tries to test things that depend on the feature being preffed off, right?

Comment 17

[(no longer active)]

(In reply to comment #16)
> That said, the only problem arises if you have a test that takes a preffed-off
> feature, flips the pref, tests the feature, flips the pref back, and then tries
> to test things that depend on the feature being preffed off, right?

Oh wait.  Comment 14 was per window (as in, the global window object), right?  In that case, please ignore what I said!

Comment 18

[Boris Zbarsky [:bzbarsky]]

Yes, per inner window.  So as soon as you work with a different window, or navigate to a different page, you will get the pref check again.

Comment 19

[(no longer active)]

(In reply to comment #18)
> Yes, per inner window.  So as soon as you work with a different window, or
> navigate to a different page, you will get the pref check again.

Perfect!