You are viewing a plain text version of this content. The canonical link for it is here.
Posted to cactus-dev@jakarta.apache.org by Xuan Nguyen <xn...@it.swin.edu.au> on 2005/07/09 16:32:50 UTC

Cactus & Cargo Integration

Hi everyone,
Thank Vincent for the suggestion in the last post. I have decided to keep
the Cactus tasks. From the user's side, running cactus test remains with 3
steps (start container/cactify/config+start+stop). So my plan is rather
straight forward: Modifying Cactus Task so it: 
1.call the cargo ant task to set the war/ear in configuration, start the
container up. 
2.start the test  (junit) 
3. call the cargo ant task to shutdown the container

The start/config/shutdown will be modified through containerset task.

I suspect about an absolute transparent integration (unless someone can
convince me this :). In my opinion, we need to pass argument like
containerid, port, action, etc...into Cargo side and hence a change on these
attributes from Cargo side will have impact on Cactus.
I would be appreciate for having ideas from your guys on this soon

Thanks
Xuan


---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


Re: Cactus & Cargo Integration

Posted by Felipe Leme <ma...@felipeal.net>.
Vincent Massol wrote:

> Why not. However, I'd be in favor of implementing approach 1 first and do 2
> later on.

Ok, agreed.

> In any case, I don't disagree! :-) I just think it's more important to
> provide 1 first and I'm just wondering whether we have the time/energy to
> maintain the 2 approaches. But if there are Cactus committers willing to
> support the 2 approaches, I think it's fine!

Well, we will need to support something on the Maven plugin anyways...


-- Felipe

---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Vincent Massol <vm...@pivolis.com>.

> -----Original Message-----
> From: Felipe Leme [mailto:maven@felipeal.net]
> Sent: mardi 12 juillet 2005 01:40
> To: Cactus Developers List
> Subject: Re: Cactus & Cargo Integration
> 
> Vincent Massol wrote:
> 
> > The goal of this Cargo refactoring is really to remove all aspects of
> > container configuration and manipulation from Cactus. Thus it should be
> > possible for someone to use his/her own ways of doing this or to use
> Cargo
> > which should integrate well.
> 
> I agree. In particular, sometimes you need to run the test in a
> container that is already running (and shouldn't be started again by
> Cacctus).
> 
> > That said, I don't see much advantage of having that and I think I would
> > prefer the 5 steps above.
> >
> > What do other think?
> 
> I'm not sure if I understood it, but if I did, you mean there are too
> possible ways to use Cactus integrated with Cargo (on Ant):
> 
> 1.User must define all (5) steps on build.xml
> 2.We supply a <cactus> tag that does everything.
> 
> I particularly think we should offer both options, i.e., a single task
> that does everything based on properties and another task that lets the
> user define what to do.

Why not. However, I'd be in favor of implementing approach 1 first and do 2
later on.
 
> Also, note that #2 is the way that Maven works, so if we move that
> funcionality to a proper Ant task, the maven plugin would be much simpler.

Ant is about providing small tasks to users so that he/she can integrate
them in his/her own build. They are small building blocks.

Maven's approach is completely different. It does not provide building
block: it provides full-fledged build features.

In any case, I don't disagree! :-) I just think it's more important to
provide 1 first and I'm just wondering whether we have the time/energy to
maintain the 2 approaches. But if there are Cactus committers willing to
support the 2 approaches, I think it's fine!

[snip]

Thanks
-Vincent


---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


Re: Cactus & Cargo Integration

Posted by Felipe Leme <ma...@felipeal.net>.
Vincent Massol wrote:

> The goal of this Cargo refactoring is really to remove all aspects of
> container configuration and manipulation from Cactus. Thus it should be
> possible for someone to use his/her own ways of doing this or to use Cargo
> which should integrate well.

I agree. In particular, sometimes you need to run the test in a 
container that is already running (and shouldn't be started again by 
Cacctus).

> That said, I don't see much advantage of having that and I think I would
> prefer the 5 steps above.
> 
> What do other think?

I'm not sure if I understood it, but if I did, you mean there are too 
possible ways to use Cactus integrated with Cargo (on Ant):

1.User must define all (5) steps on build.xml
2.We supply a <cactus> tag that does everything.

I particularly think we should offer both options, i.e., a single task 
that does everything based on properties and another task that lets the 
user define what to do.

Also, note that #2 is the way that Maven works, so if we move that 
funcionality to a proper Ant task, the maven plugin would be much simpler.

-- Felipe

PS: I will try to fix the last bugs before the 1.7.1 release tonight and 
then create the CVS branch for it...

---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


Re: Cactus & Cargo Integration

Posted by Felipe Leme <ma...@felipeal.net>.
Vincent,

I had applied Xuan's last patches last week, but haven't committed the 
changes (don't quite remember why; I think I had a question and later 
realized the answer myself, so Xuan didn't reply).

Anyway, my point is that I will be too busy in the next weeks so I'd 
like to ask you to take care of this patch (and the final ones) - could 
you please do that?

Tks,

-- Felipe

Xuan Thang Nguyen wrote:
> Sorry that I forgot to attach the build file in the last email. It is
> modified in the execute-tests.

---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Xuan Thang Nguyen <xn...@it.swin.edu.au>.
Hi Vincent,
Sorry that I forgot to attach the build file in the last email. It is
modified in the execute-tests.
Thanks
Xuan

-----Original Message-----
From: Xuan Thang Nguyen [mailto:xnguyen@it.swin.edu.au] 
Sent: Wednesday, 13 July 2005 11:49 AM
To: 'Cactus Developers List'
Subject: RE: Cactus & Cargo Integration 


Hi Vincent, sorry for the confusion.
Step 4, instead of using a junit task, I create a new tag "PerformTest", the
rest 4 steps are kept unchanged. PerformTestTask extends JUnitTask. 
For an example of how PerformTest tag looks like, in reference to the
build.xml you attach in the previous mail. 

<PerformTest 
 
contextURL="http://localhost:${cactus.port}/${project.name.file}-cactified"
  testreportDir="${target.testreports.dir}/@{containerKey}"
  log ="${target.logfile}">

  <classpath>
    <path refid="cactus.classpath"/>
    <pathelement location="${build.classes.dir}"/>
    <pathelement location="${test.classes.dir}"/>
  </classpath>

  <formatter type="xml"/>
  <batchtest>
    <fileset dir="${src.test.dir}">
      <include name="**/Test*.java"/>
      <exclude name="**/TestAll.java"/>
    </fileset>
  </batchtest>
</PerformTest>

I put this in option 2 even there is a "new" tag since it's only a
convinient wrapping of junit. Thanks Xuan




---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Xuan Thang Nguyen <xn...@it.swin.edu.au>.
Hi Vincent, sorry for the confusion.
Step 4, instead of using a junit task, I create a new tag "PerformTest", the
rest 4 steps are kept unchanged. PerformTestTask extends JUnitTask. 
For an example of how PerformTest tag looks like, in reference to the
build.xml you attach in the previous mail. 

<PerformTest 
 
contextURL="http://localhost:${cactus.port}/${project.name.file}-cactified"
  testreportDir="${target.testreports.dir}/@{containerKey}"
  log ="${target.logfile}">

  <classpath>
    <path refid="cactus.classpath"/>
    <pathelement location="${build.classes.dir}"/>
    <pathelement location="${test.classes.dir}"/>
  </classpath>

  <formatter type="xml"/>
  <batchtest>
    <fileset dir="${src.test.dir}">
      <include name="**/Test*.java"/>
      <exclude name="**/TestAll.java"/>
    </fileset>
  </batchtest>
</PerformTest>

I put this in option 2 even there is a "new" tag since it's only a
convinient wrapping of junit.
Thanks
Xuan




---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Vincent Massol <vm...@pivolis.com>.
Hi Xuan,

Sorry but I'm having a hard time trying to visualize what you mean. Could
you please show an example using the XML tags?

What is option 2? Is it about wrapping the following in a task:

  <macrodef name="junit-container">
    <attribute name="containerKey"/>
    <sequential>
      <mkdir dir="${target.testreports.dir}"/>
      <mkdir dir="${target.testreports.dir}/@{containerKey}"/>
      <property prefix="cactus.log" 
          file="${target.dir.normalized}/logging_client.properties"/>
      <junit fork="yes" failureproperty="tests.failed" haltonerror="true">
        <sysproperty key="cactus.contextURL" 
 
value="http://localhost:${cactus.port}/${project.name.file}-cactified"/>
        <syspropertyset>
          <propertyref prefix="cactus.log"/>
          <mapper type="glob" from="cactus.log.*" to="*"/>
        </syspropertyset>
        <classpath>
          <path refid="project.classpath"/>
          <path refid="cactus.classpath"/>
          <@clover.begin@pathelement location="${clover.jar}"/@clover.end@>
          <pathelement location="${httpunit.jar}"/>
          <pathelement location="${nekohtml.jar}"/>
          <pathelement location="${target.classes.java.dir}"/>
          <pathelement location="${target.classes.cactus.dir}"/>
          <pathelement location="${log4j.jar}"/>
        </classpath>
        <formatter type="brief" usefile="false"/>
        <formatter type="xml"/>
        <batchtest todir="${target.testreports.dir}/@{containerKey}">
          <fileset dir="${src.cactus.dir}">
            <!-- Due to some Cactus synchronization bug, the 'unit' tests
                 Need to run before the 'sample' tests -->
            <include name="**/servlet/unit/Test*.java"/>
            <exclude name="**/servlet/unit/Test*All.java"/>
          </fileset>
        </batchtest>
        <batchtest todir="${target.testreports.dir}/@{containerKey}">
          <fileset dir="${src.cactus.dir}">
            <include name="**/servlet/Test*.java"/>
            <exclude name="**/servlet/Test*All.java"/>
          </fileset>
        </batchtest>
      </junit>
    </sequential>
  </macrodef>

?

(this comes from the CARGO branch)

This is how a user would use the <junit> task directly to run Cactus tests.
I think the lightweight wrapping task could make it easier by:
- setting the sysproperty properties
- setting the classpath

The rest should be left to the user.

Are we on the same wavelength?

Thanks
-Vincent

> -----Original Message-----
> From: Xuan Thang Nguyen [mailto:xnguyen@it.swin.edu.au]
> Sent: dimanche 10 juillet 2005 12:49
> To: 'Cactus Developers List'
> Subject: RE: Cactus & Cargo Integration
> 
> Hi Vincent,
> Since no one has had a comments so far. I assume it's safe to go further
> for
> option 2 in which  let's create a new attribute "performTest" similar to
> runservertests as an extension of junit. It will have the following tags:
> - testURL	To check the server has been started up
> - timeout	To stop listening if no response from the server after a
> period of time
> - testreportDir	To hold the generated reports
> - log		Cactus log file
> 
> Apparently there is nothing new, these are collected from Cactus &
> runservertests tasks.
> 
> If you are happy with this, I'll go ahead.

[snip]

RE: Cactus & Cargo Integration

Posted by Xuan Thang Nguyen <xn...@it.swin.edu.au>.
Hi Vincent,
Since no one has had a comments so far. I assume it's safe to go further for
option 2 in which  let's create a new attribute "performTest" similar to
runservertests as an extension of junit. It will have the following tags:
- testURL	To check the server has been started up
- timeout	To stop listening if no response from the server after a
period of time
- testreportDir	To hold the generated reports
- log		Cactus log file

Apparently there is nothing new, these are collected from Cactus &
runservertests tasks.

If you are happy with this, I'll go ahead.

Thanks
Xuan


---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Vincent Massol <vm...@pivolis.com>.
It's fine with me Xuan.

Thanks
-Vincent

> -----Original Message-----
> From: Xuan Nguyen [mailto:xnguyen@it.swin.edu.au]
> Sent: dimanche 10 juillet 2005 04:02
> To: 'Cactus Developers List'
> Subject: RE: Cactus & Cargo Integration
> 
> Hi everyone,
> To summarize, we have three options here
> 1. Hide Cargo completely from users (my original approach)
> 2. Let users to use Cargo attributes (start/config/stop)without
> introducing
> any new attributes(Vincent first approach)
> 3. Let users to use Cargo with new nested attributes which use Cargo
> (Vincent second approach)
> 
> For me 2 and 3 are similar, 3 is more clean. However, I also prefer 2 as I
> don't see much use in 3.
> As the final purpose is for a transparent integration, I vote for  2
> though
> it gives the users some extra work and their knowledge on cargo.
> 
> Any one has other options for this and what do you think ?
> 
> I would like to have a final agreement on this matter soon while
> understanding that your guys may be busy for other stuff ;)
> 
> Cheers
> Xuan
> 
> -----Original Message-----
> From: Vincent Massol [mailto:vmassol@pivolis.com]
> Sent: Sunday, 10 July 2005 2:14 AM
> To: 'Cactus Developers List'
> Subject: RE: Cactus & Cargo Integration
> 
> 
> Hi Xuan,
> 
> > -----Original Message-----
> > From: Xuan Nguyen [mailto:xnguyen@it.swin.edu.au]
> > Sent: samedi 9 juillet 2005 16:33
> > To: 'Cactus Developers List'
> > Subject: Cactus & Cargo Integration
> >
> > Hi everyone,
> > Thank Vincent for the suggestion in the last post. I have decided to
> > keep the Cactus tasks. From the user's side, running cactus test
> > remains with 3 steps (start container/cactify/config+start+stop). So
> > my plan is rather straight forward: Modifying Cactus Task so it:
> > 1.call the cargo ant task to set the war/ear in configuration, start
> > the container up. 2.start the test  (junit)
> > 3. call the cargo ant task to shutdown the container
> >
> > The start/config/shutdown will be modified through containerset task.
> >
> > I suspect about an absolute transparent integration (unless someone
> > can convince me this :). In my opinion, we need to pass argument like
> > containerid, port, action, etc...into Cargo side and hence a change on
> > these attributes from Cargo side will have impact on Cactus.
> > I would be appreciate for having ideas from your guys on this soon
> 
> While one task is good from a usage point of view, I suspect it would
> force
> us to be too much tied to Cargo.
> 
> The goal of this Cargo refactoring is really to remove all aspects of
> container configuration and manipulation from Cactus. Thus it should be
> possible for someone to use his/her own ways of doing this or to use Cargo
> which should integrate well.
> 
> If you keep on master cactus task, I suspect you'll need to tie Cactus to
> Cargo in some manner and this would make us more or less dependent on
> Cargo,
> with the need to redeliver a new version whenever Cargo changes.
> 
> Now to answer your question, no there's no need for Cactus to pass any
> argument like containerid, port, action to the Cargo tasks. This is done
> by
> the user and that's the beauty of it! :-)
> 
> All in all, I think I would prefer about having leaving the user to call
> the
> Cargo tasks to configure/start/stop containers and to have a Cactus task
> wrapping the <junit> Ant task to make it easy for users (it would set
> cactus
> properties, set up cactus logging, etc). Actually this could be very
> similar
> to the existing runservertests Ant task (if not the same!).
> 
> Are you able to get this independence with a single Ant task? Could you
> show
> us an example of your task in action so that we can see what you mean?
> 
> Once again what I mean is a 5 steps:
> 
> 1/ Create the WAR/EAR using the <war> task for exemple
> 2/ Cactify the WAR/EAR using the <cactifywar>, <cactifyear> tasks 3/
> Configure and start the container using the Cargo task. E.g. <cargo-
> resin3x
> [...] action="start"/> 4/ Use a <junit> wrapper to start the tests 5/ Stop
> the container using the Cargo tasks E.g. <cargo-resin3x action="stop"/>
> 
> Now, what I mean by the runservertests task is that it's possible to have
> something like this (combines steps 3 to 5 from above):
> 
> <cactus war|ear="...">
>   <start>
>     <cargo-resin3x id="mycontainer" [...] action="start"/>
>   </start>
>   <stop>
>     <cargo-resin3x refid="mycontainer" [...] action="stop"/>
>   </stop>
> </cactus>
> 
> That would mean that the user would be free to use whatever Ant tasks he
> wants between the <start/> and <stop/> tags.
> 
> That said, I don't see much advantage of having that and I think I would
> prefer the 5 steps above.
> 
> What do other think?
> 
> Thanks
> -Vincent
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: cactus-dev-help@jakarta.apache.org
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: cactus-dev-help@jakarta.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Xuan Nguyen <xn...@it.swin.edu.au>.
Hi everyone,
To summarize, we have three options here
1. Hide Cargo completely from users (my original approach)
2. Let users to use Cargo attributes (start/config/stop)without introducing
any new attributes(Vincent first approach)
3. Let users to use Cargo with new nested attributes which use Cargo
(Vincent second approach)

For me 2 and 3 are similar, 3 is more clean. However, I also prefer 2 as I
don't see much use in 3.
As the final purpose is for a transparent integration, I vote for  2 though
it gives the users some extra work and their knowledge on cargo.

Any one has other options for this and what do you think ?

I would like to have a final agreement on this matter soon while
understanding that your guys may be busy for other stuff ;)

Cheers
Xuan 

-----Original Message-----
From: Vincent Massol [mailto:vmassol@pivolis.com] 
Sent: Sunday, 10 July 2005 2:14 AM
To: 'Cactus Developers List'
Subject: RE: Cactus & Cargo Integration 


Hi Xuan,

> -----Original Message-----
> From: Xuan Nguyen [mailto:xnguyen@it.swin.edu.au]
> Sent: samedi 9 juillet 2005 16:33
> To: 'Cactus Developers List'
> Subject: Cactus & Cargo Integration
> 
> Hi everyone,
> Thank Vincent for the suggestion in the last post. I have decided to 
> keep the Cactus tasks. From the user's side, running cactus test 
> remains with 3 steps (start container/cactify/config+start+stop). So 
> my plan is rather straight forward: Modifying Cactus Task so it: 
> 1.call the cargo ant task to set the war/ear in configuration, start 
> the container up. 2.start the test  (junit)
> 3. call the cargo ant task to shutdown the container
> 
> The start/config/shutdown will be modified through containerset task.
> 
> I suspect about an absolute transparent integration (unless someone 
> can convince me this :). In my opinion, we need to pass argument like 
> containerid, port, action, etc...into Cargo side and hence a change on 
> these attributes from Cargo side will have impact on Cactus.
> I would be appreciate for having ideas from your guys on this soon

While one task is good from a usage point of view, I suspect it would force
us to be too much tied to Cargo.

The goal of this Cargo refactoring is really to remove all aspects of
container configuration and manipulation from Cactus. Thus it should be
possible for someone to use his/her own ways of doing this or to use Cargo
which should integrate well.

If you keep on master cactus task, I suspect you'll need to tie Cactus to
Cargo in some manner and this would make us more or less dependent on Cargo,
with the need to redeliver a new version whenever Cargo changes.

Now to answer your question, no there's no need for Cactus to pass any
argument like containerid, port, action to the Cargo tasks. This is done by
the user and that's the beauty of it! :-)

All in all, I think I would prefer about having leaving the user to call the
Cargo tasks to configure/start/stop containers and to have a Cactus task
wrapping the <junit> Ant task to make it easy for users (it would set cactus
properties, set up cactus logging, etc). Actually this could be very similar
to the existing runservertests Ant task (if not the same!).

Are you able to get this independence with a single Ant task? Could you show
us an example of your task in action so that we can see what you mean?

Once again what I mean is a 5 steps:

1/ Create the WAR/EAR using the <war> task for exemple
2/ Cactify the WAR/EAR using the <cactifywar>, <cactifyear> tasks 3/
Configure and start the container using the Cargo task. E.g. <cargo-resin3x
[...] action="start"/> 4/ Use a <junit> wrapper to start the tests 5/ Stop
the container using the Cargo tasks E.g. <cargo-resin3x action="stop"/>

Now, what I mean by the runservertests task is that it's possible to have
something like this (combines steps 3 to 5 from above):

<cactus war|ear="...">
  <start>
    <cargo-resin3x id="mycontainer" [...] action="start"/>
  </start>
  <stop>
    <cargo-resin3x refid="mycontainer" [...] action="stop"/>
  </stop>
</cactus>

That would mean that the user would be free to use whatever Ant tasks he
wants between the <start/> and <stop/> tags.

That said, I don't see much advantage of having that and I think I would
prefer the 5 steps above.

What do other think?

Thanks
-Vincent


---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org


RE: Cactus & Cargo Integration

Posted by Vincent Massol <vm...@pivolis.com>.
Hi Xuan,

> -----Original Message-----
> From: Xuan Nguyen [mailto:xnguyen@it.swin.edu.au]
> Sent: samedi 9 juillet 2005 16:33
> To: 'Cactus Developers List'
> Subject: Cactus & Cargo Integration
> 
> Hi everyone,
> Thank Vincent for the suggestion in the last post. I have decided to keep
> the Cactus tasks. From the user's side, running cactus test remains with 3
> steps (start container/cactify/config+start+stop). So my plan is rather
> straight forward: Modifying Cactus Task so it:
> 1.call the cargo ant task to set the war/ear in configuration, start the
> container up.
> 2.start the test  (junit)
> 3. call the cargo ant task to shutdown the container
> 
> The start/config/shutdown will be modified through containerset task.
> 
> I suspect about an absolute transparent integration (unless someone can
> convince me this :). In my opinion, we need to pass argument like
> containerid, port, action, etc...into Cargo side and hence a change on
> these
> attributes from Cargo side will have impact on Cactus.
> I would be appreciate for having ideas from your guys on this soon

While one task is good from a usage point of view, I suspect it would force
us to be too much tied to Cargo.

The goal of this Cargo refactoring is really to remove all aspects of
container configuration and manipulation from Cactus. Thus it should be
possible for someone to use his/her own ways of doing this or to use Cargo
which should integrate well.

If you keep on master cactus task, I suspect you'll need to tie Cactus to
Cargo in some manner and this would make us more or less dependent on Cargo,
with the need to redeliver a new version whenever Cargo changes.

Now to answer your question, no there's no need for Cactus to pass any
argument like containerid, port, action to the Cargo tasks. This is done by
the user and that's the beauty of it! :-)

All in all, I think I would prefer about having leaving the user to call the
Cargo tasks to configure/start/stop containers and to have a Cactus task
wrapping the <junit> Ant task to make it easy for users (it would set cactus
properties, set up cactus logging, etc). Actually this could be very similar
to the existing runservertests Ant task (if not the same!).

Are you able to get this independence with a single Ant task? Could you show
us an example of your task in action so that we can see what you mean?

Once again what I mean is a 5 steps:

1/ Create the WAR/EAR using the <war> task for exemple
2/ Cactify the WAR/EAR using the <cactifywar>, <cactifyear> tasks
3/ Configure and start the container using the Cargo task. E.g.
<cargo-resin3x [...] action="start"/>
4/ Use a <junit> wrapper to start the tests
5/ Stop the container using the Cargo tasks E.g. <cargo-resin3x
action="stop"/>

Now, what I mean by the runservertests task is that it's possible to have
something like this (combines steps 3 to 5 from above):

<cactus war|ear="...">
  <start>
    <cargo-resin3x id="mycontainer" [...] action="start"/>
  </start>
  <stop>
    <cargo-resin3x refid="mycontainer" [...] action="stop"/>
  </stop>
</cactus>

That would mean that the user would be free to use whatever Ant tasks he
wants between the <start/> and <stop/> tags.

That said, I don't see much advantage of having that and I think I would
prefer the 5 steps above.

What do other think?

Thanks
-Vincent


---------------------------------------------------------------------
To unsubscribe, e-mail: cactus-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: cactus-dev-help@jakarta.apache.org