jsr344-experts@javaserverfaces-spec-public.java.net

[jsr344-experts] Re: Re: ClientWindow API review

From: Andy Schwartz <andy.schwartz_at_oracle.com>
Date: Wed, 13 Mar 2013 23:18:49 -0400

Hi Ed -

On 3/13/13 11:07 PM, Edward Burns wrote:
> AS> The ClientWindow's parameter map should be immutable.
>
> I have added a spec statement saying the map must be immutable.

Excellent, thanks!

> I
> didn't go as far as requiring an exception be thrown if it turns out not
> to be immutable, however.
>

Sure. Specifying that the returned Map is immutable is sufficient.

>>> Also, for performance I think
>>> checking for a null return is quicker than an empty map.
>>>
>
> AS> This is not a good approach to performance optimization.
>
> I have changed it to return the empty map, not null.
>

Great.

> AS> Also, should we spec that when disabled,
> AS> ClientWindow.getQueryURLParameters() returns the empty collection? If
> AS> we did that, do we even need an isClientWindowRenderModeEnabled()
> AS> API?
>
> Again, I'm going to push back on that one.
>

That's fine. I am less worried about this specific API than about
building up some foundation of design principles that the
expert/volunteer group can share for our future work (2.3 and beyond).
We can take this up against at the start of 2.3.

> AS> When it comes to API design, I am a minimalist. [...] As a
> AS> minimalist, I take pleasure in killing off seemingly pointless APIs.
> AS> :-)
>
> This is why I accepted and implemented your suggestions about decode()
> and attachWindow().
>

Great, thanks again Ed.

Andy