I would prefer DS but if we could agree on dependency manager then that
would still be better than our custom code.
Btw. I would be interested to hear some more about your experiences and
problems with DS. Till now I got most things running fine.
Maybe we can chat on irc or google hangouts and then recap what we found
here.
Talking about Java 8 .. karaf 4.1 now requires java 8 so we should be
able to use the new features.
Christian
On 18.03.2016 10:50, Łukasz Dywicki wrote:
-1 With current codebase I must say that I can not support this idea. DS is
more limited than blueprint and custom activators we currently have. I already
tried to wire some things with DS and it is not possible - ie. making JAAS
realm with login modules inside is not possible. Starting thread from it or
wiring in bundle context requires hand crafting or wrappers. Far closer to what
I would appreciate is Felix Dependency Manager [1][2], which will have similar
frames as our org.apache.karaf.util (base activator class) but better syntax
and chances for a bit wider community of users. There is also fresh module
which supports Java 8 syntax which is much more flexible than creating
components by hand [3]. Some day, if we will switch to Java 8 it will be
blessing, for now it is just an extra module.
For sure felix dependency manager is less popular than declarative services
which are part of osgi spec, however I’m sure it will have even smaller memory
footprint than DS cause it is still plain Java which does not require XML
parsing, will be faster to start up, and will offer similar or smaller
complexity (see troubles with more advanced injections in DS).
I must say that when I saw new Activators in Karaf 4 I was a bit lost, but at
the end it wasn’t that hard to pick it up and use karaf.util in my own stuff.
What I do miss for community is lack of documentation and example projects.
More over in some places it is similar to Felix Dependency Manager which makes
me wonder if we could use it/join forces instead of reinventing the wheel. For
sure it is extra dependency which may conflict with users, but may also
interest people who use Felix DM to come and use Karaf. :-)
[1]
http://felix.apache.org/documentation/subprojects/apache-felix-dependency-manager.html
<http://felix.apache.org/documentation/subprojects/apache-felix-dependency-manager.html>
[2] https://github.com/apache/felix/tree/trunk/dependencymanager
<https://github.com/apache/felix/tree/trunk/dependencymanager>
[3]
https://github.com/apache/felix/tree/trunk/dependencymanager/org.apache.felix.dependencymanager.lambda.samples/src/org/apache/felix/dm/lambda/samples/hello
<https://github.com/apache/felix/tree/trunk/dependencymanager/org.apache.felix.dependencymanager.lambda.samples/src/org/apache/felix/dm/lambda/samples/hello>
Best regards,
Łukasz Dywicki
--
Apache Karaf Committer & PMC
Twitter: ldywicki
Blog: http://dywicki.pl
Code-House - http://code-house.org
Wiadomość napisana przez Jean-Baptiste Onofré <[email protected]> w dniu 17 mar
2016, o godz. 20:15:
-0
I'm puzzled here:
- on one hand, I see that it can help for development, improve support of scr
for some parts of Karaf (shell commands, etc), so it could be a good move
- on the other hand, it's a core dependency, as we had for blueprint, so with
the same problem: even the minimal distribution will require DS/SCR support.
Anyway, if we go this way, it will be on a first major release (maybe 5.x as we
removed blueprint from 3.x to 4.x).
Regards
JB
On 03/17/2016 04:43 PM, Christian Schneider wrote:
We currently use some custom Activator base classes to wire the karaf
bundles. The goal of this was to avoid depending on blueprint
as it is a quite heavy dependency and makes it harder to use a different
blueprint impl or version.
There are some problems with this approach though:
- It makes it harder for new people to understand what we are doing
- The custom code is more error prone than a proven framework
So I propose to switch our own bundles to use DS to expose and wire
services.
There are some advantages:
- The DS annotation approach is easier to understand and more self
documenting than the custom code
- We get rid of the classes in util for the custom code
- The scr commands help diagnose problems
The main cost is that we need to always install the felix scr bundle.
To prove that it can work I switched bundle core in a branch
https://github.com/apache/karaf/tree/EXPERIMENTAL_DS .
The DS based code works quite nicely.
Btw. I found a small problem with our shell command extender. It only
seems to work on all commands or none. If there is any required service
missing then none of the commands is installed.
This made it hard for me to diagnose problems as I was missing all
bundle commands ;-)
So while working on the switch I thought about two improvements to the
extender:
1. Work on each command individually. So each command can activate as
soon as the deps are met
2. Provide a service and commands to diagnose problems like the scr
commands
Christian
--
Jean-Baptiste Onofré
[email protected] <mailto:[email protected]>
http://blog.nanthrax.net <http://blog.nanthrax.net/>
Talend - http://www.talend.com <http://www.talend.com/>
--
Christian Schneider
http://www.liquid-reality.de
Open Source Architect
http://www.talend.com
