Re: Quartz, next

2013-12-12 Thread AndyG
I see what you mean now - As shade only works at the 'package' phase there is
no way round this other than to use your approach - An external project
dependency. Dammit
 

Andy



--
View this message in context: 
http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p4666717.html
Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-12 Thread AndyG
You need to enable the 'main' profile in the maven projects view, then the
whole world comes in to say hello (takes a while even on a fast machine).
Also enable automatic imports in the intellij maven settings.

Andy.



--
View this message in context: 
http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p4666718.html
Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-11 Thread AndyG
So how would we feel with 'openejb-shade-[quartz]' ? And eventually bring all
shade jars in line with that naming?

I'm happy to do the maven stuff either way.

Andy



--
View this message in context: 
http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p486.html
Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-11 Thread AndyG
I agree. Just checked in for a build something that works for me - We can
rename stuff later.

Andy.



--
View this message in context: 
http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p488.html
Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-11 Thread Romain Manni-Bucau
Hmm

I have not enough time to say no but till try to justify my opinion (i like
product-openejb-shade (or shaded))

It really makes diff easy and if product and shade are together by mistake
in assemblies you see it quickly sorting by name. Last point for libraries
which were not shaded in previous versions it lets the artifact at the same
place which makes the transition smoother and avoids issues in pom,
assembly.xml, exclusions

Now technically all works ;)
Le 11 déc. 2013 10:54, AndyG andy.gumbre...@orprovision.com a écrit :

 I agree. Just checked in for a build something that works for me - We can
 rename stuff later.

 Andy.



 --
 View this message in context:
 http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p488.html
 Sent from the OpenEJB Dev mailing list archive at Nabble.com.



Re: Quartz, next

2013-12-11 Thread David Blevins

On Dec 10, 2013, at 2:40 PM, David Blevins david.blev...@gmail.com wrote:

 As far as where the shaded project lives, I'm cool with whatever as long as 
 it doesn't:
 - break Intellij so we can still compile with a plain Intellij import and not 
 having to go through any magic to compile/run

Looks like the trunk/shade/quartz/ setup is no different than the former 
trunk/deps/ setup.  Breaks the IDE so you can't compile or run tests.


-David



Re: Quartz, next

2013-12-10 Thread AndyG
Damn, was too late checking this in.

The only issue with quartz-openejb-shade was trying to use the quartz
version rather than 'omitting' the version so that it resolves to the parent
project version (i.e. 4.6.1-SNAPSHOT) - This is how it should be, as we
really want to keep control of the shade at build time when we deploy to
either a snapshot or release repo.

Moving it out of the project completely is IMHO not the way to go as it
complicates the build/deploy and release even more.

Andy.



--
View this message in context: 
http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p449.html
Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-10 Thread Romain Manni-Bucau
+1 to keep it in the build (in old deps)

however on the renaming I think keeping the shade starting with
quartz makes it more obvious.
Romain Manni-Bucau
Twitter: @rmannibucau
Blog: http://rmannibucau.wordpress.com/
LinkedIn: http://fr.linkedin.com/in/rmannibucau
Github: https://github.com/rmannibucau



2013/12/10 AndyG andy.gumbre...@orprovision.com:
 And renaming to openejb-quartz-shade would make sense i.e. produce
 openejb-quartz-shade-4.6.1-SNAPSHOT.jar



 --
 View this message in context: 
 http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p451.html
 Sent from the OpenEJB Dev mailing list archive at Nabble.com.


Re: Quartz, next

2013-12-10 Thread Romain Manni-Bucau
Exactly why i would like to avoid it, ls quartz*. That's what is expected
(from me at least). This is smoother to compare with released versions IMO
Le 10 déc. 2013 17:43, AndyG andy.gumbre...@orprovision.com a écrit :

 The name 'openejb-quartz-shade' just keeps it nice in a directory listing
 (i.e. All openejb jars together)



 --
 View this message in context:
 http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p456.html
 Sent from the OpenEJB Dev mailing list archive at Nabble.com.



Re: Quartz, next

2013-12-10 Thread David Blevins
I suspect what Romain is pointing out is we already have things like 
openejb-derby, openejb-hsql and openejb-cxf.  No way at this point to use 
openejb-cxf to imply a shaded or patched version of CXF without it 
conflicting with the existing openejb-cxf module where the CXF integration code 
lives.

As far as where the shaded project lives, I'm cool with whatever as long as it 
doesn't:
 - break Intellij so we can still compile with a plain Intellij import and not 
having to go through any magic to compile/run
 - break the maven-assembly-plugin and pull the binaries twice
 - require us to ping/pong the shaded module in and out of the build

We used to have the shaded javaee-api jar in the build, but we only pulled it 
in when one of the jars needed to be patched.  The result was we were randomly 
adding it and then on those releases we'd need to remember to kill it 
afterwards and it actually made the release process way harder -- one person 
adding something to the build that some other person would have to remember to 
clean up.  No fun.

In or out, I'm fine with either as long as we stick to it.


-David

On Dec 10, 2013, at 10:15 AM, Romain Manni-Bucau rmannibu...@gmail.com wrote:

 Exactly why i would like to avoid it, ls quartz*. That's what is expected
 (from me at least). This is smoother to compare with released versions IMO
 Le 10 déc. 2013 17:43, AndyG andy.gumbre...@orprovision.com a écrit :
 
 The name 'openejb-quartz-shade' just keeps it nice in a directory listing
 (i.e. All openejb jars together)
 
 
 
 --
 View this message in context:
 http://openejb.979440.n4.nabble.com/Quartz-next-tp4666498p456.html
 Sent from the OpenEJB Dev mailing list archive at Nabble.com.
 



Re: Quartz, next

2013-12-05 Thread Mark Struberg


+1 for shading. Always good to prevent classpath clashes without having to 
provide expensive ClassLoader tricks.

The only obstacle is if there is some configuration which needs to be read from 
a file with a fixed name.

LieGrue,
strub






 From: David Blevins david.blev...@gmail.com
To: dev@tomee.apache.org 
Sent: Thursday, 5 December 2013, 2:38
Subject: Re: Quartz, next
 

Trying the shaded approach would certainly be interesting.

There's a lot of merit to it if we can work through the obvious issues and 
test the heck out of it.


-David


On Dec 3, 2013, at 12:57 AM, Romain Manni-Bucau rmannibu...@gmail.com wrote:

 Ps: forgot another solutuon: shade quartz in org.apache.openejb.quartz
 Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com a écrit :
 
 Hi
 
 How do we handle quartz for next releases? It is very often a pain
 
 I propose:
 1) if in openejb loader use this one
 2) if not look for it in tomee/quartz/*.jar and create a loader with it
 
 Tomee would use 2 by default
 
 Wdyt?
 





Re: Quartz, next

2013-12-05 Thread Romain Manni-Bucau
hopefully it should work now. I didnt check tomee only provides quartz
shade but our quartz ra tests are passing (if any volonteer...)

Note: I kept org.quartz config in EjbTimerServiceImpl but we should
replace it by org.apache.openejb.quartz.* (here again if any volonteer
;).
Romain Manni-Bucau
Twitter: @rmannibucau
Blog: http://rmannibucau.wordpress.com/
LinkedIn: http://fr.linkedin.com/in/rmannibucau
Github: https://github.com/rmannibucau



2013/12/5 Mark Struberg strub...@yahoo.de:


 +1 for shading. Always good to prevent classpath clashes without having to 
 provide expensive ClassLoader tricks.

 The only obstacle is if there is some configuration which needs to be read 
 from a file with a fixed name.

 LieGrue,
 strub






 From: David Blevins david.blev...@gmail.com
To: dev@tomee.apache.org
Sent: Thursday, 5 December 2013, 2:38
Subject: Re: Quartz, next


Trying the shaded approach would certainly be interesting.

There's a lot of merit to it if we can work through the obvious issues and 
test the heck out of it.


-David


On Dec 3, 2013, at 12:57 AM, Romain Manni-Bucau rmannibu...@gmail.com wrote:

 Ps: forgot another solutuon: shade quartz in org.apache.openejb.quartz
 Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com a écrit :

 Hi

 How do we handle quartz for next releases? It is very often a pain

 I propose:
 1) if in openejb loader use this one
 2) if not look for it in tomee/quartz/*.jar and create a loader with it

 Tomee would use 2 by default

 Wdyt?






Re: Quartz, next

2013-12-04 Thread Jean-Louis MONTEIRO
But, the main objective of shading is to let users embed their own Quartz
if I understood without interfering with TomEE internals.
So if they embed it, where is the issue? They gonna be able to
extend/implement whatever they want to.

JLouis


2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com

 quartz api (interfaces/abstract classes) are sometimes used by users.
 If we shade we can't use default impl. Not blocking for me, was mainly
 a warning
 Romain Manni-Bucau
 Twitter: @rmannibucau
 Blog: http://rmannibucau.wordpress.com/
 LinkedIn: http://fr.linkedin.com/in/rmannibucau
 Github: https://github.com/rmannibucau



 2013/12/3 Jean-Louis MONTEIRO jeano...@gmail.com:
  Did not get time to dig into, was just my first feeling. Could you
  elaborate a bit more?
  Not sure I understood the extension issue?
 
  JLouis
 
 
  2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com
 
  The 3rd ;). The issue is we loose all quartz extensions then
  Le 3 déc. 2013 10:10, Jean-Louis MONTEIRO jeano...@gmail.com a
 écrit :
 
   Maybe the second (shaded) one is easier to manage at first glance.
   So that is my preferred solution
  
  
   2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com
  
Ps: forgot another solutuon: shade quartz in
 org.apache.openejb.quartz
Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com
 a
écrit :
   
 Hi

 How do we handle quartz for next releases? It is very often a pain

 I propose:
 1) if in openejb loader use this one
 2) if not look for it in tomee/quartz/*.jar and create a loader
 with
  it

 Tomee would use 2 by default

 Wdyt?

   
  
  
  
   --
   Jean-Louis
  
 
 
 
 
  --
  Jean-Louis




-- 
Jean-Louis


Re: Quartz, next

2013-12-04 Thread Romain Manni-Bucau
But no more the tomee one without copying code. (once again not blocking)
Le 4 déc. 2013 09:35, Jean-Louis MONTEIRO jeano...@gmail.com a écrit :

 But, the main objective of shading is to let users embed their own Quartz
 if I understood without interfering with TomEE internals.
 So if they embed it, where is the issue? They gonna be able to
 extend/implement whatever they want to.

 JLouis


 2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com

  quartz api (interfaces/abstract classes) are sometimes used by users.
  If we shade we can't use default impl. Not blocking for me, was mainly
  a warning
  Romain Manni-Bucau
  Twitter: @rmannibucau
  Blog: http://rmannibucau.wordpress.com/
  LinkedIn: http://fr.linkedin.com/in/rmannibucau
  Github: https://github.com/rmannibucau
 
 
 
  2013/12/3 Jean-Louis MONTEIRO jeano...@gmail.com:
   Did not get time to dig into, was just my first feeling. Could you
   elaborate a bit more?
   Not sure I understood the extension issue?
  
   JLouis
  
  
   2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com
  
   The 3rd ;). The issue is we loose all quartz extensions then
   Le 3 déc. 2013 10:10, Jean-Louis MONTEIRO jeano...@gmail.com a
  écrit :
  
Maybe the second (shaded) one is easier to manage at first glance.
So that is my preferred solution
   
   
2013/12/3 Romain Manni-Bucau rmannibu...@gmail.com
   
 Ps: forgot another solutuon: shade quartz in
  org.apache.openejb.quartz
 Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com
 
  a
 écrit :

  Hi
 
  How do we handle quartz for next releases? It is very often a
 pain
 
  I propose:
  1) if in openejb loader use this one
  2) if not look for it in tomee/quartz/*.jar and create a loader
  with
   it
 
  Tomee would use 2 by default
 
  Wdyt?
 

   
   
   
--
Jean-Louis
   
  
  
  
  
   --
   Jean-Louis
 



 --
 Jean-Louis



Re: Quartz, next

2013-12-04 Thread David Blevins
Trying the shaded approach would certainly be interesting.

There's a lot of merit to it if we can work through the obvious issues and test 
the heck out of it.


-David

On Dec 3, 2013, at 12:57 AM, Romain Manni-Bucau rmannibu...@gmail.com wrote:

 Ps: forgot another solutuon: shade quartz in org.apache.openejb.quartz
 Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com a écrit :
 
 Hi
 
 How do we handle quartz for next releases? It is very often a pain
 
 I propose:
 1) if in openejb loader use this one
 2) if not look for it in tomee/quartz/*.jar and create a loader with it
 
 Tomee would use 2 by default
 
 Wdyt?
 



Re: Quartz, next

2013-12-03 Thread Romain Manni-Bucau
Ps: forgot another solutuon: shade quartz in org.apache.openejb.quartz
Le 2 déc. 2013 22:23, Romain Manni-Bucau rmannibu...@gmail.com a écrit :

 Hi

 How do we handle quartz for next releases? It is very often a pain

 I propose:
 1) if in openejb loader use this one
 2) if not look for it in tomee/quartz/*.jar and create a loader with it

 Tomee would use 2 by default

 Wdyt?