You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by Robert Munteanu <ro...@apache.org> on 2018/11/09 12:41:02 UTC
[feature model] Designating a feature as 'complete'
Hi,
I am wondering if it makes sense to mark a feature as 'complete'. A
complete feature would be one that is expected to be launched
individually, e.g. needs to additions to function.
I would see the benefits mainly in tooling:
- a complete feature is self-contained, therefore all requirements must
be satisfied
- a non-complete feature may not be launched, so don't try to do that
Thoughts?
Robert
Re: [feature model] Designating a feature as 'complete'
Posted by Carsten Ziegeler <cz...@apache.org>.
I've created https://issues.apache.org/jira/browse/SLING-8100 and marked
it as a blocker for the next release
Regards
Carsten
Am 13.11.2018 um 08:02 schrieb Carsten Ziegeler:
>
>
> Am 09.11.2018 um 14:23 schrieb Robert Munteanu:
>> On Fri, 2018-11-09 at 13:49 +0100, Carsten Ziegeler wrote:
>>> - what exactly does complete imply? can it be launched with other
>>> features? etc. I guess we just need to define this.
>>
>> My personal optinion is that we should be able to launch a complete
>> feature with other features. One scenario which would be supported is ,
>> having a complete feature which can be launched standalone and then
>> adding instrumentation/monitoring on top of that.
>>
>> One thing I am not sure about is whether there should be any barriers
>> preventing two complete features from being launched together. At first
>> glance it does not make sense, but on the other hand people find
>> creative uses which were not originally taken into account :-)
>>
> Right, there is also the question if you assemble a complete and a non
> complete feature, is the result complete? (I guess the answer is no)
>
> So there are some smaller things to find out.
>
> As we're close to finalize the feature API, I think we should create an
> issue for this and make sure we include it as early as possible
>
> Regards
> Carsten
>
--
Carsten Ziegeler
Adobe Research Switzerland
cziegeler@apache.org
Re: [feature model] Designating a feature as 'complete'
Posted by Carsten Ziegeler <cz...@apache.org>.
Am 09.11.2018 um 14:23 schrieb Robert Munteanu:
> On Fri, 2018-11-09 at 13:49 +0100, Carsten Ziegeler wrote:
>> - what exactly does complete imply? can it be launched with other
>> features? etc. I guess we just need to define this.
>
> My personal optinion is that we should be able to launch a complete
> feature with other features. One scenario which would be supported is ,
> having a complete feature which can be launched standalone and then
> adding instrumentation/monitoring on top of that.
>
> One thing I am not sure about is whether there should be any barriers
> preventing two complete features from being launched together. At first
> glance it does not make sense, but on the other hand people find
> creative uses which were not originally taken into account :-)
>
Right, there is also the question if you assemble a complete and a non
complete feature, is the result complete? (I guess the answer is no)
So there are some smaller things to find out.
As we're close to finalize the feature API, I think we should create an
issue for this and make sure we include it as early as possible
Regards
Carsten
--
Carsten Ziegeler
Adobe Research Switzerland
cziegeler@apache.org
Re: [feature model] Designating a feature as 'complete'
Posted by Robert Munteanu <ro...@apache.org>.
On Fri, 2018-11-09 at 13:49 +0100, Carsten Ziegeler wrote:
> - what exactly does complete imply? can it be launched with other
> features? etc. I guess we just need to define this.
My personal optinion is that we should be able to launch a complete
feature with other features. One scenario which would be supported is ,
having a complete feature which can be launched standalone and then
adding instrumentation/monitoring on top of that.
One thing I am not sure about is whether there should be any barriers
preventing two complete features from being launched together. At first
glance it does not make sense, but on the other hand people find
creative uses which were not originally taken into account :-)
Robert
Re: [feature model] Designating a feature as 'complete'
Posted by Robert Munteanu <ro...@apache.org>.
On Fri, 2018-11-09 at 13:12 +0000, David Bosschaert wrote:
> FWIW the current description of features vs 'complete features' in
> the OSGi
> RFP can be found in section 2.4 of [1].
>
> @Robert Munteanu <ro...@apache.org> pretty much what you described
> :)
Ah OK. I should've read that first :-)
Thanks,
Robert
>
> Best regards,
>
> David
>
> [1]
> https://github.com/osgi/design/blob/master/rfps/rfp-0188-Features.pdf
>
> On Fri, 9 Nov 2018 at 12:49, Carsten Ziegeler <cz...@apache.org>
> wrote:
>
> > We had a similar discussion in the OSGi expert group about this and
> > basically came to a similar conclusion.
> >
> > Now, in the first version of our code base in Sling we
> > distinguished
> > between a feature and an application, where an application was a
> > complete feature. We dropped the application concept as it was not
> > really adding something new. The only thing really needed is some
> > kind
> > of a marker as you suggest.
> >
> > There are some aspects to consider:
> > - framework launch properties: while a feature can have framework
> > properties, only a complete feature can define framework launch
> > properties. Not sure if we have to model something here or just
> > hope
> > that people do the right thing when defining their features
> > - what exactly does complete imply? can it be launched with other
> > features? etc. I guess we just need to define this.
> >
> > Regards
> > Carsten
> >
> >
> > Am 09.11.2018 um 13:41 schrieb Robert Munteanu:
> > > Hi,
> > >
> > > I am wondering if it makes sense to mark a feature as 'complete'.
> > > A
> > > complete feature would be one that is expected to be launched
> > > individually, e.g. needs to additions to function.
> > >
> > > I would see the benefits mainly in tooling:
> > >
> > > - a complete feature is self-contained, therefore all
> > > requirements must
> > > be satisfied
> > > - a non-complete feature may not be launched, so don't try to do
> > > that
> > >
> > > Thoughts?
> > >
> > > Robert
> > >
> >
> > --
> > Carsten Ziegeler
> > Adobe Research Switzerland
> > cziegeler@apache.org
> >
Re: [feature model] Designating a feature as 'complete'
Posted by David Bosschaert <da...@gmail.com>.
FWIW the current description of features vs 'complete features' in the OSGi
RFP can be found in section 2.4 of [1].
@Robert Munteanu <ro...@apache.org> pretty much what you described :)
Best regards,
David
[1] https://github.com/osgi/design/blob/master/rfps/rfp-0188-Features.pdf
On Fri, 9 Nov 2018 at 12:49, Carsten Ziegeler <cz...@apache.org> wrote:
> We had a similar discussion in the OSGi expert group about this and
> basically came to a similar conclusion.
>
> Now, in the first version of our code base in Sling we distinguished
> between a feature and an application, where an application was a
> complete feature. We dropped the application concept as it was not
> really adding something new. The only thing really needed is some kind
> of a marker as you suggest.
>
> There are some aspects to consider:
> - framework launch properties: while a feature can have framework
> properties, only a complete feature can define framework launch
> properties. Not sure if we have to model something here or just hope
> that people do the right thing when defining their features
> - what exactly does complete imply? can it be launched with other
> features? etc. I guess we just need to define this.
>
> Regards
> Carsten
>
>
> Am 09.11.2018 um 13:41 schrieb Robert Munteanu:
> > Hi,
> >
> > I am wondering if it makes sense to mark a feature as 'complete'. A
> > complete feature would be one that is expected to be launched
> > individually, e.g. needs to additions to function.
> >
> > I would see the benefits mainly in tooling:
> >
> > - a complete feature is self-contained, therefore all requirements must
> > be satisfied
> > - a non-complete feature may not be launched, so don't try to do that
> >
> > Thoughts?
> >
> > Robert
> >
>
> --
> Carsten Ziegeler
> Adobe Research Switzerland
> cziegeler@apache.org
>
Re: [feature model] Designating a feature as 'complete'
Posted by Carsten Ziegeler <cz...@apache.org>.
We had a similar discussion in the OSGi expert group about this and
basically came to a similar conclusion.
Now, in the first version of our code base in Sling we distinguished
between a feature and an application, where an application was a
complete feature. We dropped the application concept as it was not
really adding something new. The only thing really needed is some kind
of a marker as you suggest.
There are some aspects to consider:
- framework launch properties: while a feature can have framework
properties, only a complete feature can define framework launch
properties. Not sure if we have to model something here or just hope
that people do the right thing when defining their features
- what exactly does complete imply? can it be launched with other
features? etc. I guess we just need to define this.
Regards
Carsten
Am 09.11.2018 um 13:41 schrieb Robert Munteanu:
> Hi,
>
> I am wondering if it makes sense to mark a feature as 'complete'. A
> complete feature would be one that is expected to be launched
> individually, e.g. needs to additions to function.
>
> I would see the benefits mainly in tooling:
>
> - a complete feature is self-contained, therefore all requirements must
> be satisfied
> - a non-complete feature may not be launched, so don't try to do that
>
> Thoughts?
>
> Robert
>
--
Carsten Ziegeler
Adobe Research Switzerland
cziegeler@apache.org