Just in case you don not need all the fancy magic of pax exam what is
really only rarely needed in most cases you can use osgi-test framework
as a lightweight alternative:
https://github.com/osgi/osgi-test/
sadly their maven example is a bit overcrowded, you can find a more
focused example
OSGi spec is on the way to move to jakarta and noticed that there is
actually no one in the OSGI-WG to drive the web stuff and if it makes
sense to retire these parts from further specification releases[1].
As I know pax-web (and karaf as well) is quite active in this area I'd
like to invite
Well obviously log4j being an ASF project has not protected it from
beeing affected by one of the worst bugs, neither has "solarwind" who
most probably knows all developers in person so I don't get it either,
this all for me is more a false-security feeling or just a generic "it
might be
I can only encourage everyone that get "complains" or "concerns" of "big
bussiness" or even single users telling them to simply start
contribution or funding OS projects they depend on:
participation/review/testing (especially upcoming versions) is the best
way to mitigate
Hi all,
I'd like to let you know that I have now used pax-jpa for some year ins
several projects and it does its job quite well.
For that reason I'm currently upgrading dependencies, cleaning up some
code and then like to release a 1.0.0 version of it.
I also plan to add some test-cases
I think one problem might be that some of the projects would then be
bound to karaf even more as they are currently are because karaf
"drives" them.
Also it might make feel people that they should only be used with karaf
and are not standalone.
So for me it won't make any sense to name them
Sounds like a good idea.
To be honest, I would even okay with NOT migrating to much of the
issues. There are a lot of "dead" ones and as long as we can preserve
the JIRA ones as a reference a link might suffice as well.
Am 01.02.21 um 09:38 schrieb Grzegorz Grzybek:
Hello
Some time ago[1]
:
Hello
śr., 18 lis 2020 o 08:23 'Christoph Läubrich' via OPS4J
mailto:ops4j@googlegroups.com>> napisał(a):
Not sure if it already was mentioned, but OSGi provides already a
way to
interact with Java-SPI through the Service Loader Mediator
Specification
[1].
'Christoph Läubrich' via OPS4J
mailto:ops4j@googlegroups.com>> napisał(a):
Not sure if it already was mentioned, but OSGi provides already a
way to
interact with Java-SPI through the Service Loader Mediator
Specification
[1].
[1]
https://docs.osgi.org/specifi
Not sure if it already was mentioned, but OSGi provides already a way to
interact with Java-SPI through the Service Loader Mediator Specification
[1].
[1] https://docs.osgi.org/specification/osgi.cmpn/7.0.0/service.loader.html
Am 16.11.20 um 09:47 schrieb Grzegorz Grzybek:
Hello! See my
Just keep in mind that a while back there was a discussion about pax
exam 5 because there was work from harald on a branch and the work on
the master branch.
Even though I tried to merge most of that stuff into master if I rember
right some tests are failing.
So be prepared that there are
+1 from me as we essentilally haven't used the real power of jira over
the past years and it was a little bit brittle to register for two services
Am 27.04.20 um 09:57 schrieb Grzegorz Grzybek:
+1 - especially for commit-issue references!
--
--
--
OPS4J - http://www.ops4j.org
ct we don't have any staging on
releases could be seen as an issue.
That's why it could be interesting to have this under the "Karaf Umbrella".
Regards
JB
On Thu, Feb 20, 2020 at 7:58 AM 'Christoph Läubrich' via OPS4J
mailto:ops4j@googlegroups.com>> wrote:
I always enjoyed
I always enjoyed the ease of contribution via github with minimal
effort. While for apache-projects has always felt like a mess because of
different hurdles.
So I would defiantly vote for staying with github, dropping JIRA in
favour of github issues (they improved the issue/project handling a
Hi Grzegorz,
I struggeled with that long time ago also and it was the reason to move
away from pax-web and using the directly the jetty-OSGI-integration
directly, from my point of view, pax-web is to much centric about
"internal convers" aka making it work for the white-boardextender, but
I already commented there, in a perfect world there would be no need to
have extra libs to use JDBC drivers in OSGi, thats why I add support for
it whenever possible even though it often takes some time (approx 8
month in both cases):
https://github.com/microsoft/mssql-jdbc/pull/700
This is currently only avaiable in the snapshot-build, I'm not sure if
they are avaiable again, so you might want to build a snapshot on your
local machine,
I'm currently preparing to extract the code into a seperate repro but
this might takes some time, but you can still contact me if you
WHat is teh reason for merging the META-INF at all?
Beside that why embeeed log4j at all if it is already an OSGi Bundle?
I think that people do embeed a way to much stuff in the OSGi world that
does not make sense. Embedding should only be used if the lib is not an
OSGi-Jar (where I prefere
Can you elaborate a little bit more what you want to archive? Starting
an OSGi-Framework is nothing special and does not require pax-runner in
any way, pax runner was used to configure the launch, but that is very
specific, for example configuring the launch would be more a task of
eclipse IDE
/SQLServerDataSourceFactory.java
Am 01.05.19 um 16:49 schrieb Steinar Bang:
'Christoph Läubrich' via OPS4J :
DriverManger.register/unregister can't work reliable in OSGi
(especially with static insitilizer), thus always the
DataSourceFactory should be used instead!
PGDataSourceFactory is currently
DriverManger.register/unregister can't work reliable in OSGi (especially
with static insitilizer), thus always the DataSourceFactory should be
used instead!
Am 30.04.19 um 19:43 schrieb Steinar Bang:
Christian Schneider :
I see no calls to DriverManager in
null in the next major release
Am 12.10.18 um 09:11 schrieb Grzegorz Grzybek:
pt., 12 paź 2018 o 08:56 'Christoph Läubrich' via OPS4J
mailto:ops4j@googlegroups.com>> napisał(a):
Just wondering if it is okay for BND if the interface has a (Java 8)
default implemen
Just wondering if it is okay for BND if the interface has a (Java 8)
default implementation (if this is possible), that should be compatible
with previous release.
Am 12.10.18 um 08:45 schrieb Grzegorz Grzybek:
Hello Kal,
I tried cherry-picking your change to pax-web-7.2.x branch, but as I
Yes thats a good idea.
Am 21.09.18 um 06:39 schrieb nino martinez wael:
Hi Guys
Wicket 8.1 have been released, and the current feature for PAX Wicket
5.0.0.RC1 WICKET are 8.0.0.m7 should we get it bumped up and make a
propper release of pax wicket.. We've spent some time testing it, and it
Take a look at servicetracker or better jet use DeclarativeSerices for
that kind of stuff.
Waiting for a service inside an activator is waiting for bad things
happen ;-)
Am 11.09.2018 um 19:10 schrieb Nhut Thai Le:
Hello,
I'm trying to create a shared httpContext so i can use it across
I would consider using the systembundles bundle context is bad style!
This would break instantly the whole code in a secured OSGi env or would
require to give the extender huge access rights.
Besdie this, it make IMO the subsystem aproach useless, it was meant to
seperate things, if you now
What about simply adding them to your test as a bundle? :-)
I think Pax Exam should not include/install ANYTHING by default because
this always creates problems with the test-setup!
I'm currently working on a way where the test-probe is the only thing
that gets installed so there are no
This strategy (PerSuite) is only supported in Java EE, CDI and Web modes.
Also please read the warning about "PerSuite" here:
https://ops4j1.jira.com/wiki/spaces/PAXEXAM4/pages/54263848/Reactor+Strategies
so it is very likely that it would produce inconsistent behaviour in
OSGi modes.
Am
What should PaxExam Inject if the component does not publish a service?
What should it check for? All this can better be handled bei a JUnit
rule as described before.
Am 23.02.2018 um 13:55 schrieb Oliver Lietz:
On Wednesday 21 February 2018 13:47:46 'Christoph Läubrich' via OPS4J wrote
Well I think the best would be to have a JUnit rule that uses the SCR
Service to query for a component.
This rule then can fail if component is not present at all or wait up to
x seconds and so on... If I find some time the next days I'll try to
prepare an example for this.
--
--
some kind of Whitebord-pattern or notifications.
If you describe your usecase a little bit more one might suggest a way
how it can be archived...
Am 21.02.2018 um 13:41 schrieb Oliver Lietz:
On Wednesday 21 February 2018 13:07:07 'Christoph Läubrich' via OPS4J wrote:
What exaclty do you want
is just an internal concept of SCR from the outside a
component does not look any different than other services and should be
handeled as those.
Am 21.02.2018 um 12:59 schrieb Oliver Lietz:
On Wednesday 21 February 2018 10:47:05 'Christoph Läubrich' via OPS4J wrote:
This depends on the DS-
The name is derived from the Exam-System with createID but I think that
can't be accessed from your test.
You might be able to set it by providing an own ProbeBuilder. But can't
you simply create the mapping at runtime? e.g. with a @Before you can
fetch the current Bunlde name (via injected
This depends on the DS-Provider you use e.g. the apache felix one has an
API for this and if I rember right it is also standarized in OSGi-R7
See
http://felix.apache.org/documentation/subprojects/apache-felix-service-component-runtime.html#api-use
This is nothing special in pax exam so you
-Baptiste Onofré:
Hi Christoph,
I think we are almost good for 4.12.0 release.
Regards
JB
On 11/24/2017 10:00 AM, 'Christoph Läubrich' via OPS4J wrote:
Hi,
I'd like to know what is left fro the 4.12.0 relase?
I created a pull-request[1] for the Eclipse-Container feature (merged
from the current
i'm currently working on docu+tests for the eclipse container, maybe we
can wait for this to include it?
Am 13.02.2018 um 06:09 schrieb Jean-Baptiste Onofré:
No yet, but no objection from my side.
Regards
JB
On 02/12/2018 11:57 PM, Oliver Lietz wrote:
hi all,
are there any release plans
Okay, do you think we should try to clean-up a bit now the new
JIRA/Clunfluence versions are out?
I think the bes would be to have the follwoing strucutre:
- anonymous: can read all JIRA + Confluence
- guest: assigned to all new registered users as default role) can also
only read, so new ppl
While working on the codebase of Pax Exam the last month an idea cam
into my mind I'd like to discuss.
Over the years Pax Exam evolved from an "OSGi-Container-Starter" to a
general-purpose container-testing-framework with lots of features and
container-types (J2EE, CDi, OSGi, ...) but also
documentation on how to use the new container
and what the use cases really are.
IMHO both tests and docs are a prerequisite for including this in any
release, be it 4.x or 5.x.
Regards,
Harald
Am 24.11.2017 um 10:07 schrieb 'Christoph Läubrich' via OPS4J:
That sounds great, do you think you
Hi,
I'd like to know what is left fro the 4.12.0 relase?
I created a pull-request[1] for the Eclipse-Container feature (merged
from the current 5.x development), since this just adds a new container
there won't be any side effects, but I wonder if there is any work
on-going before we can
If I view https://ops4j1.jira.com/browse/PAXEXAM-832 while logged in all
is fine, stories show up and the link to the Wiki is displayed.
If I'm not logged in (e.g. open a private tab, view it on a different
computer) the Stories are not displayed and the link to the wiki states
that I'm not
Am 12.11.2017 um 18:51 schrieb Harald Wellmann:
Now how can we consolidate all that ongoing work?
Hi Harald what do you think about the following procedure:
- we create 1 branches of the current master (exam5-ongoing) to capture
current work for reference
- create an exam5-integration branch
I also think this is definately not a micro-change! If it is
incompatible with client-code then this has to be a major version bump.
Beside this I think embedding things should only be used in very rare
cases but instead we should try to use bnds class check to make sure we
always use a
builder to bnd
.. but this would create compatibility problems.
Christian
2017-10-05 9:58 GMT+02:00 'Christoph Läubrich' via OPS4J
<ops4j@googlegroups.com <mailto:ops4j@googlegroups.com>>:
> I only use this in the @Configuration so my test should not require
> Tiny
It all depends of the mode you rune your tests (mebedded or forked)
pax-exma is pulling nothing "in" or exposing anything to OSGi if you do
not ask for it.
You should also distinguisch between maven-depndencies declared in your
pom and maven-urls declared in feature files.
The first are
Hi,
currently the process of starting a "remote-container" is very
implementation-specific and requires special code.
For Pax Exam 5 I'm currently reworking the handling of (Test-)Containers
and will then provide a way for more reliable and generic run and use
remote containers so this
> I only use this in the @Configuration so my test should not require
> TinyBundles at runtime.
Even though you does not require the code to run in @Configure you still
have an (class-import) dependecy to that code.
Is it a real problem to include tiny-bundles in your test?
If yes you have to
Hi,
I mean that you already have the datasource at hand so there is no need
to look it up in JNDI. But it seems there is no way of connecting camel
with the underlying persitence unit.
BTW: Your problem has nothing to do with pax-jdbc or pax-exam it is more
related to aries-jpa that does
From the log it seems your blueprint starts up before the service is
registered. Since Camel does not know directly about the dependency you
might try the following:
Let your blueprint import a javax.sql.DataSource service, then it won't
start up before one is avaiable e.g. with
that will
Okay I have checked this and it does not seems a real option. With a
custom cunstructor wen can't use standard ServcieLoader but must
implement our own, and we can't declare checked exceptions in a nice way.
Beside this, at least the Karaf-Container does some magic in its
Factory, so I think we
Hi Nino,
I just pushed a Wicket-8 supporting version of pax-wicket
(https://ops4j1.jira.com/browse/PAXWICKET-486) to the master, the only
thing that is needed is adjusting the karaf-test, can you help with
that? I think it just needs to add some of the new dependencies, I added
the following
I'd like to start a new OPS4j pax project that is about remote control
of OSGi, maybe it will implements (but not restricted to) "Remote
Services" and "Remote Service Admin" Specification.
The main goal for now would be to provide support for new container
abstraction in pax-exam but I think
I suspect that pax-jdbc is NOT reading *cfg files directly but you
provide them through felix-fileinstall as regular OSGi-Configurations
and pax-jdbc use osgi-cm to access them.
You have two options:
- hook up on felix-fileinstall by providing a custom ArtifactTransformer
and
Hi Toni,
If i see right there is some kind of snapshot repro for OPS4j maybe that
would be an Option? I'm not very familar with this I just have seen that
when realeasing pax-wicket.
If thats not an option we might provide some kind of preview1 ... x
Versions?
Am 08.08.2017 09:53, schrieb
Currently all new issues are assigend by default to Harald, I think we
should change this to "unassiged" so everyone can pick up the issue in
question.
With the actual default is is a bit confusing since it is unclear if
assigned to Harald he is activly working (or plan to work) on it or it
Even though neo4j is not "osgi-friendly" (what ever that means ;-) you
can create abundle that embedds all that neo4j requires, add an
Activator that starts up your db and provide a custom service to trigger
"after-test-checks" in whatever way you need. In your test you can the
collect the
Ah okay, its been a while I tried BNDTools :-)
I created an Epic to track all work related to BNDTools [1] so you might
want to add more things as the came up.
I think it would be good to support as much platforms as possible, I'm
currently workin in integrating Eclipse/PDE to work more nicly
Hi Toni,
I have a list of ideas here I try to bring to the JIRA the next days :-)
Primarly these are related to allow more than one container to be
started inside test (e.g. for client/server tests) and having a pertest
choice of the container to run as well as making COnfiguration of
Can you explain what is the problem with bnd<->exam? Even though I don't
use Bndtool atm I'm interessted in it.
Shouldn't BNDTools work with any OSGi bundle?
You might want to take a look at the bnd-maven plugin
http://njbartlett.name/2015/03/27/announcing-bnd-maven-plugin.html that
allos to
I encountered a problem and have no clue what causing this, see
https://ops4j1.jira.com/browse/PAXEXAM-841
anyone has an idea?
--
--
--
OPS4J - http://www.ops4j.org - ops4j@googlegroups.com
---
You received this message because you are subscribed to the Google Groups "OPS4J"
Hi Achim,
can you direct me to a link? I'll then check that and try to fix it, I'm
running 'mvn clean install' locally and that passes at least at two of
my machines.
What I see is that when running a full build from time to time maven
coplains about a "JVM exit without goodby", running the
We are happy to annonce the release of Pax Wicket 4 that now adds
support for Wicket 7.
You find all changes here
https://ops4j1.jira.com/wiki/display/paxwicket/Pax+Wicket+-+4.0.0
--
--
--
OPS4J - http://www.ops4j.org - ops4j@googlegroups.com
---
You received this message
I created the branch and merged all commits from the exam5-milestone,
resolved the conflicts and everything now builds fine, all tests are
running.
I'll assign PAXEXAM-829 to Harald for review.
--
--
--
OPS4J - http://www.ops4j.org - ops4j@googlegroups.com
---
You received
have hwo these should work.
Am 12.07.2017 10:58, schrieb Oliver Lietz:
On Wednesday 12 July 2017 10:24:23 'Christoph Läubrich' via OPS4J wrote:
Hi,
Hi,
I'd like to clean up the the Container API (see PAXEXAM-822 and linked
ones), simplyfy and unify the creation of OSGi-Conta
Hi,
I'd like to clean up the the Container API (see PAXEXAM-822 and linked
ones), simplyfy and unify the creation of OSGi-Container (like native,
karaf, forked) as well as adding a new Eclipse-Platform Container that
allows to work with Eclipse-Apllication and several PDE Artifacts like
the above things,
mainly pooling + XA support, but it will have to be done in a slighly
incompatible way I think. Thoughts ?
I'll start working on a branch, as I'll need a way to expose the
ConnectionFactory anyway.
2017-06-20 10:15 GMT+02:00 'Christoph Läubrich' via OPS4J
<ops4j@googlegroups.
The test are running fine now :-)
I'll try to prepare a release of a 4.0.0 Version for Wicket 7 then we
can move on to wicket 8 as a next step
Am 28.06.2017 15:21, schrieb nino martinez wael:
just made a single commit..
--
--
--
OPS4J - http://www.ops4j.org -
Didn't know we have an example/test for it :-D
I'll take a look on this on friday...
Am 28.06.2017 15:19, schrieb nino martinez wael:
I do have an problem on:
http://localhost:8181/blueprint/wiquery/
Wiquery seems to be discontinued. And it only supports wicket 7..
Maybe we should abandon
?
On Tue, Jun 20, 2017 at 10:03 AM, 'Christoph Läubrich' via OPS4J
<ops4j@googlegroups.com <mailto:ops4j@googlegroups.com>> wrote:
As long as there are no API-breaks within Pax-Wicket it doesen't
matter, client-code has to choose of course if the want 7 or 8,
but I ca
I already started https://github.com/ops4j/org.ops4j.pax.jms for JMS a
while ago so you are maybe interested in adding the pooling-feature
there also?
In fact pooling and recovery can be nicly integrated into OSGi using the
service factory, so a switch over would be simply register/unregister
:
Ahh I see most of the changes where from 6.x to 7.x..
But what about the Websocket stuff, as I remember it has been changed
alot in Wicket 8..?
On Mon, Jun 19, 2017 at 2:55 PM, 'Christoph Läubrich' via OPS4J
<ops4j@googlegroups.com <mailto:ops4j@googlegroups.com>> wrote:
I
Hi Nino,
I created the branch v3.1.x, it is working so far beside the karaf tests
fail with
Server returned HTTP response code: 401 for URL:
http://localhost:8181/issues
any idea what is causing this? If we have sorted this out I think we
should make this the 4.0 version and make wicket 8
Found the issue: The logmessage simply has wrong parameter order and a
missleading logger name, beside this this has no impact on the code
since we scan the classpath for pax-wicket annotations
The real cause for the waring is:
Hi nino,
sadly I have not had time to take a look, did you tried with the felix
http-service in the mean time?
Am 13.06.2017 12:27, schrieb nino martinez wael:
Time has passed again.. Im wondering if it would be okay to release an
experimental or milestone release with the felix test
I can try to take a look at it next week, just wondering ig the
WhiteboardService is needed for the test at all? Pax Wicket does only
require a simple http-service so you can even try to drop pax-web and
use the (simpler) felix-http service for this use case.
Am 24.05.2017 10:31, schrieb nino
Hi nino,
the goal should be to more test per release and not to disable existing
ones.
It is a little hard to help without exact informations, what is exacly
the problem with the test:
- What tests(s) are affected
- What fails (e.g. what bundles do not resolve etc.)
- Whats your
You can also take a look at OSGi Declarative Services for this task that
handles all the service tracking stuff for you...
Am 26.04.2017 11:03, schrieb nino martinez wael:
Hi Gang
Im playing around with the idea of a mybatis extender, since we use
mybatis on almost every of our projects.
and wanted. If I understood more about how the
test probe is built, I would offer to work on that now. But I can do that in
the future, if it seems like that's the best plan.
---
A. Soroka
The University of Virginia Library
On Mar 8, 2017, at 4:13 AM, 'Christoph Läubrich' via
OPS4J<op
I think an option to include a resource into the test-probe is the most
generic, simple and natural way. For special and/or advanced usage
szenarios the tinybundle option can still be used.
Maybe it would even be possible to have (as an extra) some sort of
includeMavenResources() option that
The problem is that resources are not packed inside the bundle probe.
Instead af working with classpathentries etc. directly in your
test-probe, you can create a "test-resource-bundle" that includes all
resources needed for the test, and include this in your test-setup,
beside the resources,
WHat do you mean by "apparently the Wicket devs have included non valid
bundles namely" If the bundle itself is invalid (e.g. missing imports,
invalid manifest) you should open an issue at the coresponding projekt.
If the Bundles itself have no OSGI_header information you should be able
to use
In fact this is a limitation of the underlying OSGi container (equinox?
felix?) so you can try to upgrade this. I even patched an rater old
equinox 3.7 to support Java 1.8 by simply adding the profiles file from
a newer release.
Am 13.12.2016 16:20, schrieb Sebastian Łaskawiec:
Hey!
When
You should then activate the debug options of Felix SCR to see whats
going on. Can you share the "working" vs the "not working" code? What
Pax-Container do you use? Remote? Embedded? Have you tried to use
immediate="true" for your component to make sure it is not delayed?
Am 31.10.2016 01:56,
Providing services on top of each other makes them loosely-coupled and
thus more general and more roboust.
Mangling different concerns (and DataSourceCreation and Pooling are
different) makes it less flexible and forces more dependencies to users
who simply want a DataSourceFactory. And yes DSF
84 matches
Mail list logo