On 25.03.2013, at 15:18, Carsten Ziegeler <[email protected]> wrote:

> The API can't be used accidentally in that case - even if someone
> implements it, it's not used by JCR.

It's still exposed in javadocs (that are generated by bundle) and might attract 
attention.

> Seriously, I can only follow all this wrath against this partially -
> we already have a a lot of API which can be abused and no one ever
> complained there. Even if we put this into a separate bundle, someone
> who is able to find this service and implement it, will definitely
> also be able to deploy another api bundle. So by splitting it up,
> we're not gaining anything but just spreading api which belongs
> together across bundles.

Even if it was done wrong beforehand (i.e. no separate resourceprovider-api 
bundle & package), it can be done better for new things. It doesn't really hurt 
to put it in a separate bundle. And as Bertrand noted in the other thread, it 
looks like this API might still evolve and is currently still a first shot. 
Putting it in the sling api bundle is like giving it the highest knighthood in 
Sling...

Mike's proposal of separate resourcesecurity-api and resourcesecurity bundles 
sounds good and simple to me:

https://issues.apache.org/jira/browse/SLING-2698?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13612091#comment-13612091

Cheers,
Alex

Reply via email to