Re: Replace sling launchpad with karaf?

[Bertrand Delacretaz <bd...@apache.org>]

On Fri, Sep 30, 2016 at 11:19 AM, Carsten Ziegeler <cz...@apache.org> wrote:
> ...I don't know if Karaf has anything compared to crankstart, maybe there
> is or we could contribute it - which again would have a much larger
> outreach than Sling...

I agree with that but also with Ian that our provisioning model suits
Sling extremely well, it's powerful, concise and so much nicer to
write and read that the XML that Karaf (IIUC) forces you to use.

Without having looked into the details my favorite option right now
would be to switch to Karaf, assuming it's better than Crankstart, but
continue to use our provisioning model supported by an "invisible"
translator.

-Bertrand

Re: Replace sling launchpad with karaf?

[Bertrand Delacretaz <bd...@apache.org>]

Hi,

On Tue, Oct 4, 2016 at 12:00 PM, Robert Munteanu <ro...@apache.org> wrote:

...
> So here's my 'keep' list proposal (for moving to Karaf):
> - provisioning model
> - same CLI arguments ( we can't break backwards compatibility )
> - ability to consume an arbitrary version of Felix ( AFAIK Felix
> framework upgrades are tied to new Karaf versions )
> - ability to produce JAR and WAR files ...

And also the ability to retrieve dependencies either at build time
from a Maven repository, or at run time when the instance starts, via
a pluggable mechanism.

> ...Also, a big plus is that we'll be part of the larger Karaf community
> and we'll have a slightly less exotic technology stack...

Which I like a lot.

-Bertrand

Re: Replace sling launchpad with karaf?

[Robert Munteanu <ro...@apache.org>]

On Tue, 2016-10-04 at 10:02 +0200, Bertrand Delacretaz wrote:
> On Fri, Sep 30, 2016 at 11:19 AM, Carsten Ziegeler <cziegeler@apache.
> org> wrote:
> > ...I don't know if Karaf has anything compared to crankstart, maybe
> > there
> > is or we could contribute it - which again would have a much larger
> > outreach than Sling...
> 
> 
> I agree with that but also with Ian that our provisioning model suits
> Sling extremely well, it's powerful, concise and so much nicer to
> write and read that the XML that Karaf (IIUC) forces you to use.
> 
> Without having looked into the details my favorite option right now
> would be to switch to Karaf, assuming it's better than Crankstart,
> but
> continue to use our provisioning model supported by an "invisible"
> translator.

That sounds good to me.�

I would like us to clearly define what the switch entails and what do
we get to keep:

So here's my 'keep' list proposal:

- provisioning model
- same CLI arguments ( we can't break backwards compatibility )
- ability to consume an arbitrary version of Felix ( AFAIK Felix
framework upgrades are tied to new Karaf versions )
- ability to produce JAR and WAR files

And here's my understanding of what we 'get':

- OSGi framework/startlevel/bundle management
- launcher logic
- plus all the features and extensions that Karaf has and we don't

Also, a bit plus is that we'll be part of the larger Karaf community
and we'll a slightly less exotic techonology stack.

Feel free to correct these lists, but hopefully they are a good start
towards making an informed decision.

Thanks,

Robert