I fixed the logic with https://issues.apache.org/jira/browse/SLING-7696
Regards Carsten Carsten Ziegeler wrote > I just had a quick look at the ApplicationBuilder class and its really > missing the logic. Right now its processing all features one after the > other without considering that one feature might include one of the > other provided ones. > > So the logic needs to be enhanced to first look at the includes, remove > included features and then process the remaining list > > Regards > > Carsten > > > Robert Munteanu wrote >> On Wed, 2018-05-30 at 16:50 +0200, Carsten Ziegeler wrote: >>> David Bosschaert wrote> Hi Robert,> >>> >>>> On 30 May 2018 at 14:25, Robert Munteanu <[email protected]> >>>> wrote: >>>> >>>>> Hi David, >>>>> >>>>> On Wed, 2018-05-30 at 13:40 +0100, David Bosschaert wrote: >>>>>> Hi Robert, >>>>>> >>>>>> It should be possible by creating a feature that is "including" >>>>>> another >>>>>> feature and specifies a number of removals. An example can be >>>>>> found >>>>>> at [1]. >>>>>> >>>>>> Having said that, I haven't tried this myself yet, so I'm not >>>>>> 100% >>>>>> sure >>>>>> it's fully implemented right now. >>>>> >>>>> Well, let's see if it works :-) >>>>> >>>>> Two questions before I start: >>>>> >>>>> 1. Is it OK if the feature to include + remove from is also >>>>> present >>>>> under src/main/features or will this lead to it being processed >>>>> twice? >>>>> >>>> >>>> I think it should be ok, since you're creating a new feature that >>>> includes >>>> the old feature but removes things. So as long as you refer to this >>>> new >>>> feature as your feature to use it should be fine. >>>> >>> >>> It should be ok, but I think the current application builder is not >>> yet >>> working that way and might include the feature twice. >>> >>> Regards >>> Carsten >> >> For the record, I did not manage to get this working - the removed >> bundles still end up in the application. There are too many new pieces >> for me and I have trouble following what goes where. I'll try and set >> up a minimal example that fails and we can continue the discussion >> based on that. >> >> Thanks, >> >> Robert >> -- Carsten Ziegeler Adobe Research Switzerland [email protected]
