Hi Bertrand,

On Fri, Mar 21, 2014 at 6:39 AM, Bertrand Delacretaz
<[email protected]> wrote:
> Hi Justin,
>
> On Thu, Mar 20, 2014 at 7:56 PM, Justin Edelson
> <[email protected]> wrote:
>> ...This sounds mildly silly, but why don't we leave the feature in, but
>> only enable it based on the presence of a particular feature flag?...
>
> I'd much rather have that optional sling:features support in a
> separate bundle that we can move to contrib to make its status clear.

I thought it was deteremined that this *had* to go in the resource
resolver. Attempts to do it through ResourceDecorator and
ResourceAccessGate were shouted down firmly IIRC.

>
> Right now, even if one can turn of sling:features support in the
> resource resolver, the functionality is present in a core bundle so we
> have to maintain it, and the more people use it the harder it is to
> remove it.

We don't *have* to do anything. The Apache License is really clear
about the fact that there's no implied warranty. We can very easily
mark a feature as experimental and choose to remove it later. If
anything, we should be doing more, not less of this.

>
> So, to refine my proposal I suggest:
>
> -Remove the sling:features support from the resource resolver bundle
> -Remove the corresponding code from the feature flags bundle
> -Maybe later make that functionality available again in a separate
> experimental bundle
>
> Justin, would that work for you?

Not really, based on my understanding that such an experimental bundle
isn't actually possible (i.e. that sling:feature support can only be
implemented in the Resource Resolver) and would instead be a fork of
the Resource Resolver.

I still think that if we are going to have Feature Flags *at all* in
Sling, there should be a way of using them to impact resource
resolution. This is a IMHO a natural use case. I think it is confusing
for users when we have a Sling "feature" (using the term broadly)
which so obviously relates to core framework functionality and yet
isn't integrated into the framework. This happened with the Tenant API
and I really don't want to see it happen with this.

I remain -0

Regards,
Justin


>
> Is anyone else opposed to that proposal?
>
> -Bertrand

Reply via email to