Maven2 Status... again...

classic Classic list List threaded Threaded
37 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Maven2 Status... again...

Mark Derricutt
I see on the maven-users list talk of a 2.0.5 release has surfaced - whats the likely hood there'll be surerfire/testng update timed to coincide with it?  It sounded like things were starting to move along in this area...

mark

--
It was not the presence of God that formed me, but his absence which broke me.

http://www.talios.com
http://www.flickr.com/photos/talios
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Brett Porter

Hi Mark,

They're separate releases, so it should have no impact.

Jesse and I have spoken about setting up a branch of surefire that we
all can work on to get this together, though.

- Brett

On 10/02/07, Mark Derricutt <[hidden email]> wrote:

> I see on the maven-users list talk of a 2.0.5 release has surfaced - whats
> the likely hood there'll be surerfire/testng update timed to coincide with
> it?  It sounded like things were starting to move along in this area...
>
> mark
>
> --
> It was not the presence of God that formed me, but his absence which broke
> me.
>
> http://www.talios.com
>  http://www.flickr.com/photos/talios
>  >
>


--
Apache Maven - http://maven.apache.org
"Better Builds with Maven" book - http://library.mergere.com/

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Mark Derricutt
On 2/10/07, Brett Porter <[hidden email]> wrote:

They're separate releases, so it should have no impact.

True, I was just thinking it'd be great to have a nicely timed dual-release :) 

Jesse and I have spoken about setting up a branch of surefire that we
all can work on to get this together, though.

Excellent.

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Alexandru Popescu ☀

I have had just a bit of time to scan through the existing code, and I
haven't provided any feedback yet because I think the original authors
took a route that was not so easy to grasp for me (dissecting suites
into tests, etc.). Most of the compatibility problems in the past were
caused by changes in the TestNG internal API, but this was used
because the original author has tried to "bend" TestNG to work in the
way JUnit tests are working (and IMO this is not the case).

So, currently Maven2 TestNG code takes a suite and is able to
disassemble it into <test>s which become runnable pieces. This is not
correct at least from 2 reasons:
- this leads to loosing the @Before/@AfterSuite functionality
- (due to some implementation detail) the names of tests inside a
suite definition must be unique (limitation that doesn't exist in
TestNG)

Instead I would suggest that the Maven2 code to behave the way TestNG
behaves and not introduce new semantics. So, Maven2, should take
multiple suite, and probably allow separation based on these.

There are more details I have found in the code, but I guess the above
are radical enough to guarantee a discussion :-).

BR,

./alex
--
.w( the_mindstorm )p.
  TestNG co-founder
EclipseTestNG Creator


On 2/10/07, Mark Derricutt <[hidden email]> wrote:

> On 2/10/07, Brett Porter <[hidden email]> wrote:
> >
> > They're separate releases, so it should have no impact.
>
> True, I was just thinking it'd be great to have a nicely timed dual-release
> :)
> > Jesse and I have spoken about setting up a branch of surefire that we
> > all can work on to get this together, though.
>
> Excellent.
>
>
>  >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

This is good news at least. .

Wrt the junit -like behavior I remember being told something generally
along the lines of "this was done to maintain consistent behavior with
the rest of the surefire API from a user perspective". Maybe this idea
could be re-evaluated given that:

-) All TestNG related tools (eclipse / ant / idea / etc ) more or less
go through the same basic configuration steps, so trying to change how
that is done is likely going to be an uphill battle with sometimes
b0rken results. .

-) TestNG seems to be doing a good job documenting how it functions
already, so trying to cater to junit users in the area of
configuration may not provide as much benefit as originally thought..

-) As long as the plugin does "something" reasonable with little or no
configuration (ie hopefully finding / running a few tests) that would
hopefully be enough to make both sides happy as far as default
functionality goes. ..

p.s. The maven2 surefire plugin should also be processing tests based
on simple annotation groups excluding suites I hope? (this is how I
often use it at least, ie I have no testng.xml suite definition file )

On 2/10/07, Alexandru Popescu <[hidden email]> wrote:

>
> I have had just a bit of time to scan through the existing code, and I
> haven't provided any feedback yet because I think the original authors
> took a route that was not so easy to grasp for me (dissecting suites
> into tests, etc.). Most of the compatibility problems in the past were
> caused by changes in the TestNG internal API, but this was used
> because the original author has tried to "bend" TestNG to work in the
> way JUnit tests are working (and IMO this is not the case).
>
> So, currently Maven2 TestNG code takes a suite and is able to
> disassemble it into <test>s which become runnable pieces. This is not
> correct at least from 2 reasons:
> - this leads to loosing the @Before/@AfterSuite functionality
> - (due to some implementation detail) the names of tests inside a
> suite definition must be unique (limitation that doesn't exist in
> TestNG)
>
> Instead I would suggest that the Maven2 code to behave the way TestNG
> behaves and not introduce new semantics. So, Maven2, should take
> multiple suite, and probably allow separation based on these.
>
> There are more details I have found in the code, but I guess the above
> are radical enough to guarantee a discussion :-).
>
> BR,
>
> ./alex
> --
> .w( the_mindstorm )p.
>   TestNG co-founder
> EclipseTestNG Creator
<snipped>


--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Hani Suleiman

Is it fair to say that as of right now, the maven/testng integration  
isn't really ready for most people to use, due to architectural  
issues with the surefire plugin and a TestNG API that has changed  
from when the integration was first done?

On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:

>
> This is good news at least. .
>
> Wrt the junit -like behavior I remember being told something generally
> along the lines of "this was done to maintain consistent behavior with
> the rest of the surefire API from a user perspective". Maybe this idea
> could be re-evaluated given that:
>
> -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> go through the same basic configuration steps, so trying to change how
> that is done is likely going to be an uphill battle with sometimes
> b0rken results. .
>
> -) TestNG seems to be doing a good job documenting how it functions
> already, so trying to cater to junit users in the area of
> configuration may not provide as much benefit as originally thought..
>
> -) As long as the plugin does "something" reasonable with little or no
> configuration (ie hopefully finding / running a few tests) that would
> hopefully be enough to make both sides happy as far as default
> functionality goes. ..
>
> p.s. The maven2 surefire plugin should also be processing tests based
> on simple annotation groups excluding suites I hope? (this is how I
> often use it at least, ie I have no testng.xml suite definition file )
>
> On 2/10/07, Alexandru Popescu <[hidden email]>  
> wrote:
>>
>> I have had just a bit of time to scan through the existing code,  
>> and I
>> haven't provided any feedback yet because I think the original  
>> authors
>> took a route that was not so easy to grasp for me (dissecting suites
>> into tests, etc.). Most of the compatibility problems in the past  
>> were
>> caused by changes in the TestNG internal API, but this was used
>> because the original author has tried to "bend" TestNG to work in the
>> way JUnit tests are working (and IMO this is not the case).
>>
>> So, currently Maven2 TestNG code takes a suite and is able to
>> disassemble it into <test>s which become runnable pieces. This is not
>> correct at least from 2 reasons:
>> - this leads to loosing the @Before/@AfterSuite functionality
>> - (due to some implementation detail) the names of tests inside a
>> suite definition must be unique (limitation that doesn't exist in
>> TestNG)
>>
>> Instead I would suggest that the Maven2 code to behave the way TestNG
>> behaves and not introduce new semantics. So, Maven2, should take
>> multiple suite, and probably allow separation based on these.
>>
>> There are more details I have found in the code, but I guess the  
>> above
>> are radical enough to guarantee a discussion :-).
>>
>> BR,
>>
>> ./alex
>> --
>> .w( the_mindstorm )p.
>>   TestNG co-founder
>> EclipseTestNG Creator
> <snipped>
>
>
> --
> Jesse Kuhnert
> Tapestry/Dojo team member/developer
>
> Open source based consulting work centered around
> dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
>
> >


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

"Sort of" .  I can only speak to what I know works currently through
personal use..Which is a TestNG version of 5.1 and
maven-surefire-plugin version of either 2.3-SNAPSHOT or 2.8-SNAPSHOT .
(the 2.8 snapshot mostly containing fixes related to jre <= 1.4
javadoc processing as well as being able to run various forms of junit
tests )

Anything greater than TestNG 5.1 is unknown to me and likely to be broken.

As long as the maven2 stuff is being discussed, it may be a good time
to finally find a solution for getting a more automated solution for
ibiblio jar publishing. There shouldn't be any reason why Cedric
shouldn't be able to have the various jars / archives all be
automatically pushed out to the correct places with a single ant task
of some kind. We just need someone with an rsync'd repo to bootstrap
off of..

On 2/10/07, Hani Suleiman <[hidden email]> wrote:
>
> Is it fair to say that as of right now, the maven/testng integration
> isn't really ready for most people to use, due to architectural
> issues with the surefire plugin and a TestNG API that has changed
> from when the integration was first done?
>
<snipped>


--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Alexandru Popescu ☀
In reply to this post by Hani Suleiman

On 2/10/07, Hani Suleiman <[hidden email]> wrote:
>
> Is it fair to say that as of right now, the maven/testng integration
> isn't really ready for most people to use, due to architectural
> issues with the surefire plugin and a TestNG API that has changed
> from when the integration was first done?
>

Without any intentions to sound harsh, I would say that Maven/TestNG
integration problems are mainly due to the fact that the
Maven-Surefire code tried to bend TestNG to some different/unsupported
usage. And I believe that there are only 2 possible directions:

1/ have the surefire-testng code use TestNG correctly; this way we
will be able to guarantee the API and the behavior

2/ have the surefire-testng code continue the direction is started;
unfortunately I cannot say I would support this direction.

./alex
--
.w( the_mindstorm )p.
  TestNG co-founder
EclipseTestNG Creator


> On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:
>
> >
> > This is good news at least. .
> >
> > Wrt the junit -like behavior I remember being told something generally
> > along the lines of "this was done to maintain consistent behavior with
> > the rest of the surefire API from a user perspective". Maybe this idea
> > could be re-evaluated given that:
> >
> > -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> > go through the same basic configuration steps, so trying to change how
> > that is done is likely going to be an uphill battle with sometimes
> > b0rken results. .
> >
> > -) TestNG seems to be doing a good job documenting how it functions
> > already, so trying to cater to junit users in the area of
> > configuration may not provide as much benefit as originally thought..
> >
> > -) As long as the plugin does "something" reasonable with little or no
> > configuration (ie hopefully finding / running a few tests) that would
> > hopefully be enough to make both sides happy as far as default
> > functionality goes. ..
> >
> > p.s. The maven2 surefire plugin should also be processing tests based
> > on simple annotation groups excluding suites I hope? (this is how I
> > often use it at least, ie I have no testng.xml suite definition file )
> >
> > On 2/10/07, Alexandru Popescu <[hidden email]>
> > wrote:
> >>
> >> I have had just a bit of time to scan through the existing code,
> >> and I
> >> haven't provided any feedback yet because I think the original
> >> authors
> >> took a route that was not so easy to grasp for me (dissecting suites
> >> into tests, etc.). Most of the compatibility problems in the past
> >> were
> >> caused by changes in the TestNG internal API, but this was used
> >> because the original author has tried to "bend" TestNG to work in the
> >> way JUnit tests are working (and IMO this is not the case).
> >>
> >> So, currently Maven2 TestNG code takes a suite and is able to
> >> disassemble it into <test>s which become runnable pieces. This is not
> >> correct at least from 2 reasons:
> >> - this leads to loosing the @Before/@AfterSuite functionality
> >> - (due to some implementation detail) the names of tests inside a
> >> suite definition must be unique (limitation that doesn't exist in
> >> TestNG)
> >>
> >> Instead I would suggest that the Maven2 code to behave the way TestNG
> >> behaves and not introduce new semantics. So, Maven2, should take
> >> multiple suite, and probably allow separation based on these.
> >>
> >> There are more details I have found in the code, but I guess the
> >> above
> >> are radical enough to guarantee a discussion :-).
> >>
> >> BR,
> >>
> >> ./alex
> >> --
> >> .w( the_mindstorm )p.
> >>   TestNG co-founder
> >> EclipseTestNG Creator
> > <snipped>
> >
> >
> > --
> > Jesse Kuhnert
> > Tapestry/Dojo team member/developer
> >
> > Open source based consulting work centered around
> > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> >
> > >
>
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

Without knowing for sure how the maven devs feel, I would say that
moving forward with option 1 sounds like a perfectly reasonable
approach. I don't think anyone feels all decisions they make/made are
infallible in some way..

It didn't work out, move on and do it the right way so the maven2
TestNG users can start getting some sane stability.

On 2/10/07, Alexandru Popescu <[hidden email]> wrote:

>
> On 2/10/07, Hani Suleiman <[hidden email]> wrote:
> >
> > Is it fair to say that as of right now, the maven/testng integration
> > isn't really ready for most people to use, due to architectural
> > issues with the surefire plugin and a TestNG API that has changed
> > from when the integration was first done?
> >
>
> Without any intentions to sound harsh, I would say that Maven/TestNG
> integration problems are mainly due to the fact that the
> Maven-Surefire code tried to bend TestNG to some different/unsupported
> usage. And I believe that there are only 2 possible directions:
>
> 1/ have the surefire-testng code use TestNG correctly; this way we
> will be able to guarantee the API and the behavior
>
> 2/ have the surefire-testng code continue the direction is started;
> unfortunately I cannot say I would support this direction.
>
> ./alex
> --
> .w( the_mindstorm )p.
>   TestNG co-founder
> EclipseTestNG Creator
>
>
> > On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:
> >
> > >
> > > This is good news at least. .
> > >
> > > Wrt the junit -like behavior I remember being told something generally
> > > along the lines of "this was done to maintain consistent behavior with
> > > the rest of the surefire API from a user perspective". Maybe this idea
> > > could be re-evaluated given that:
> > >
> > > -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> > > go through the same basic configuration steps, so trying to change how
> > > that is done is likely going to be an uphill battle with sometimes
> > > b0rken results. .
> > >
> > > -) TestNG seems to be doing a good job documenting how it functions
> > > already, so trying to cater to junit users in the area of
> > > configuration may not provide as much benefit as originally thought..
> > >
> > > -) As long as the plugin does "something" reasonable with little or no
> > > configuration (ie hopefully finding / running a few tests) that would
> > > hopefully be enough to make both sides happy as far as default
> > > functionality goes. ..
> > >
> > > p.s. The maven2 surefire plugin should also be processing tests based
> > > on simple annotation groups excluding suites I hope? (this is how I
> > > often use it at least, ie I have no testng.xml suite definition file )
> > >
> > > On 2/10/07, Alexandru Popescu <[hidden email]>
> > > wrote:
> > >>
> > >> I have had just a bit of time to scan through the existing code,
> > >> and I
> > >> haven't provided any feedback yet because I think the original
> > >> authors
> > >> took a route that was not so easy to grasp for me (dissecting suites
> > >> into tests, etc.). Most of the compatibility problems in the past
> > >> were
> > >> caused by changes in the TestNG internal API, but this was used
> > >> because the original author has tried to "bend" TestNG to work in the
> > >> way JUnit tests are working (and IMO this is not the case).
> > >>
> > >> So, currently Maven2 TestNG code takes a suite and is able to
> > >> disassemble it into <test>s which become runnable pieces. This is not
> > >> correct at least from 2 reasons:
> > >> - this leads to loosing the @Before/@AfterSuite functionality
> > >> - (due to some implementation detail) the names of tests inside a
> > >> suite definition must be unique (limitation that doesn't exist in
> > >> TestNG)
> > >>
> > >> Instead I would suggest that the Maven2 code to behave the way TestNG
> > >> behaves and not introduce new semantics. So, Maven2, should take
> > >> multiple suite, and probably allow separation based on these.
> > >>
> > >> There are more details I have found in the code, but I guess the
> > >> above
> > >> are radical enough to guarantee a discussion :-).
> > >>
> > >> BR,
> > >>
> > >> ./alex
> > >> --
> > >> .w( the_mindstorm )p.
> > >>   TestNG co-founder
> > >> EclipseTestNG Creator
> > > <snipped>
> > >
> > >
> > > --
> > > Jesse Kuhnert
> > > Tapestry/Dojo team member/developer
> > >
> > > Open source based consulting work centered around
> > > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> > >
> > > >
> >
> >
> > >
> >
>
> >
>


--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Alexandru Popescu ☀

On 2/10/07, Jesse Kuhnert <[hidden email]> wrote:
>
> Without knowing for sure how the maven devs feel, I would say that
> moving forward with option 1 sounds like a perfectly reasonable
> approach. I don't think anyone feels all decisions they make/made are
> infallible in some way..
>
> It didn't work out, move on and do it the right way so the maven2
> TestNG users can start getting some sane stability.
>

I hope Brett Porter is watching the thread and comment on it. If not I
will ping him about it.

./alex
--
.w( the_mindstorm )p.
  TestNG co-founder
EclipseTestNG Creator

> On 2/10/07, Alexandru Popescu <[hidden email]> wrote:
> >
> > On 2/10/07, Hani Suleiman <[hidden email]> wrote:
> > >
> > > Is it fair to say that as of right now, the maven/testng integration
> > > isn't really ready for most people to use, due to architectural
> > > issues with the surefire plugin and a TestNG API that has changed
> > > from when the integration was first done?
> > >
> >
> > Without any intentions to sound harsh, I would say that Maven/TestNG
> > integration problems are mainly due to the fact that the
> > Maven-Surefire code tried to bend TestNG to some different/unsupported
> > usage. And I believe that there are only 2 possible directions:
> >
> > 1/ have the surefire-testng code use TestNG correctly; this way we
> > will be able to guarantee the API and the behavior
> >
> > 2/ have the surefire-testng code continue the direction is started;
> > unfortunately I cannot say I would support this direction.
> >
> > ./alex
> > --
> > .w( the_mindstorm )p.
> >   TestNG co-founder
> > EclipseTestNG Creator
> >
> >
> > > On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:
> > >
> > > >
> > > > This is good news at least. .
> > > >
> > > > Wrt the junit -like behavior I remember being told something generally
> > > > along the lines of "this was done to maintain consistent behavior with
> > > > the rest of the surefire API from a user perspective". Maybe this idea
> > > > could be re-evaluated given that:
> > > >
> > > > -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> > > > go through the same basic configuration steps, so trying to change how
> > > > that is done is likely going to be an uphill battle with sometimes
> > > > b0rken results. .
> > > >
> > > > -) TestNG seems to be doing a good job documenting how it functions
> > > > already, so trying to cater to junit users in the area of
> > > > configuration may not provide as much benefit as originally thought..
> > > >
> > > > -) As long as the plugin does "something" reasonable with little or no
> > > > configuration (ie hopefully finding / running a few tests) that would
> > > > hopefully be enough to make both sides happy as far as default
> > > > functionality goes. ..
> > > >
> > > > p.s. The maven2 surefire plugin should also be processing tests based
> > > > on simple annotation groups excluding suites I hope? (this is how I
> > > > often use it at least, ie I have no testng.xml suite definition file )
> > > >
> > > > On 2/10/07, Alexandru Popescu <[hidden email]>
> > > > wrote:
> > > >>
> > > >> I have had just a bit of time to scan through the existing code,
> > > >> and I
> > > >> haven't provided any feedback yet because I think the original
> > > >> authors
> > > >> took a route that was not so easy to grasp for me (dissecting suites
> > > >> into tests, etc.). Most of the compatibility problems in the past
> > > >> were
> > > >> caused by changes in the TestNG internal API, but this was used
> > > >> because the original author has tried to "bend" TestNG to work in the
> > > >> way JUnit tests are working (and IMO this is not the case).
> > > >>
> > > >> So, currently Maven2 TestNG code takes a suite and is able to
> > > >> disassemble it into <test>s which become runnable pieces. This is not
> > > >> correct at least from 2 reasons:
> > > >> - this leads to loosing the @Before/@AfterSuite functionality
> > > >> - (due to some implementation detail) the names of tests inside a
> > > >> suite definition must be unique (limitation that doesn't exist in
> > > >> TestNG)
> > > >>
> > > >> Instead I would suggest that the Maven2 code to behave the way TestNG
> > > >> behaves and not introduce new semantics. So, Maven2, should take
> > > >> multiple suite, and probably allow separation based on these.
> > > >>
> > > >> There are more details I have found in the code, but I guess the
> > > >> above
> > > >> are radical enough to guarantee a discussion :-).
> > > >>
> > > >> BR,
> > > >>
> > > >> ./alex
> > > >> --
> > > >> .w( the_mindstorm )p.
> > > >>   TestNG co-founder
> > > >> EclipseTestNG Creator
> > > > <snipped>
> > > >
> > > >
> > > > --
> > > > Jesse Kuhnert
> > > > Tapestry/Dojo team member/developer
> > > >
> > > > Open source based consulting work centered around
> > > > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> > > >
> > > > >
> > >
> > >
> > > >
> > >
> >
> > >
> >
>
>
> --
> Jesse Kuhnert
> Tapestry/Dojo team member/developer
>
> Open source based consulting work centered around
> dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

I know he is. Time zone differences may delay responses but I'm sure
they will be given.

Even more importantly, as long as the code changes mesh well with the
rest of the surefire API I'm sure it will be more well received. Ie
typical honoring of existing logic && making appropriate overall
refactorings to the whole api if they are needed instead of making one
off sort of TestNG specific only kind of logic. (which is more or less
what I originally did ) ...Just the typical un-fun keeping the whole
codebase consistent stuff...

On 2/10/07, Alexandru Popescu <[hidden email]> wrote:

>
> On 2/10/07, Jesse Kuhnert <[hidden email]> wrote:
> >
> > Without knowing for sure how the maven devs feel, I would say that
> > moving forward with option 1 sounds like a perfectly reasonable
> > approach. I don't think anyone feels all decisions they make/made are
> > infallible in some way..
> >
> > It didn't work out, move on and do it the right way so the maven2
> > TestNG users can start getting some sane stability.
> >
>
> I hope Brett Porter is watching the thread and comment on it. If not I
> will ping him about it.
>
> ./alex
> --
> .w( the_mindstorm )p.
>   TestNG co-founder
> EclipseTestNG Creator
>
> > On 2/10/07, Alexandru Popescu <[hidden email]> wrote:
> > >
> > > On 2/10/07, Hani Suleiman <[hidden email]> wrote:
> > > >
> > > > Is it fair to say that as of right now, the maven/testng integration
> > > > isn't really ready for most people to use, due to architectural
> > > > issues with the surefire plugin and a TestNG API that has changed
> > > > from when the integration was first done?
> > > >
> > >
> > > Without any intentions to sound harsh, I would say that Maven/TestNG
> > > integration problems are mainly due to the fact that the
> > > Maven-Surefire code tried to bend TestNG to some different/unsupported
> > > usage. And I believe that there are only 2 possible directions:
> > >
> > > 1/ have the surefire-testng code use TestNG correctly; this way we
> > > will be able to guarantee the API and the behavior
> > >
> > > 2/ have the surefire-testng code continue the direction is started;
> > > unfortunately I cannot say I would support this direction.
> > >
> > > ./alex
> > > --
> > > .w( the_mindstorm )p.
> > >   TestNG co-founder
> > > EclipseTestNG Creator
> > >
> > >
> > > > On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:
> > > >
> > > > >
> > > > > This is good news at least. .
> > > > >
> > > > > Wrt the junit -like behavior I remember being told something generally
> > > > > along the lines of "this was done to maintain consistent behavior with
> > > > > the rest of the surefire API from a user perspective". Maybe this idea
> > > > > could be re-evaluated given that:
> > > > >
> > > > > -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> > > > > go through the same basic configuration steps, so trying to change how
> > > > > that is done is likely going to be an uphill battle with sometimes
> > > > > b0rken results. .
> > > > >
> > > > > -) TestNG seems to be doing a good job documenting how it functions
> > > > > already, so trying to cater to junit users in the area of
> > > > > configuration may not provide as much benefit as originally thought..
> > > > >
> > > > > -) As long as the plugin does "something" reasonable with little or no
> > > > > configuration (ie hopefully finding / running a few tests) that would
> > > > > hopefully be enough to make both sides happy as far as default
> > > > > functionality goes. ..
> > > > >
> > > > > p.s. The maven2 surefire plugin should also be processing tests based
> > > > > on simple annotation groups excluding suites I hope? (this is how I
> > > > > often use it at least, ie I have no testng.xml suite definition file )
> > > > >
> > > > > On 2/10/07, Alexandru Popescu <[hidden email]>
> > > > > wrote:
> > > > >>
> > > > >> I have had just a bit of time to scan through the existing code,
> > > > >> and I
> > > > >> haven't provided any feedback yet because I think the original
> > > > >> authors
> > > > >> took a route that was not so easy to grasp for me (dissecting suites
> > > > >> into tests, etc.). Most of the compatibility problems in the past
> > > > >> were
> > > > >> caused by changes in the TestNG internal API, but this was used
> > > > >> because the original author has tried to "bend" TestNG to work in the
> > > > >> way JUnit tests are working (and IMO this is not the case).
> > > > >>
> > > > >> So, currently Maven2 TestNG code takes a suite and is able to
> > > > >> disassemble it into <test>s which become runnable pieces. This is not
> > > > >> correct at least from 2 reasons:
> > > > >> - this leads to loosing the @Before/@AfterSuite functionality
> > > > >> - (due to some implementation detail) the names of tests inside a
> > > > >> suite definition must be unique (limitation that doesn't exist in
> > > > >> TestNG)
> > > > >>
> > > > >> Instead I would suggest that the Maven2 code to behave the way TestNG
> > > > >> behaves and not introduce new semantics. So, Maven2, should take
> > > > >> multiple suite, and probably allow separation based on these.
> > > > >>
> > > > >> There are more details I have found in the code, but I guess the
> > > > >> above
> > > > >> are radical enough to guarantee a discussion :-).
> > > > >>
> > > > >> BR,
> > > > >>
> > > > >> ./alex
> > > > >> --
> > > > >> .w( the_mindstorm )p.
> > > > >>   TestNG co-founder
> > > > >> EclipseTestNG Creator
> > > > > <snipped>
> > > > >
> > > > >
> > > > > --
> > > > > Jesse Kuhnert
> > > > > Tapestry/Dojo team member/developer
> > > > >
> > > > > Open source based consulting work centered around
> > > > > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> > > > >
> > > > > >
> > > >
> > > >
> > > > >
> > > >
> > >
> > > >
> > >
> >
> >
> > --
> > Jesse Kuhnert
> > Tapestry/Dojo team member/developer
> >
> > Open source based consulting work centered around
> > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> >
> > >
> >
>
> >
>


--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

P.S. Looking back on it now , I think the overall "reason" why things
are messed up the way that they are currently can mostly be attributed
to me. Though the original set of patches I made were more along the
lines of option 1, they were done in such a hacky/inconsistent way
with the rest of the API that Brett was more or less forced to try and
take the logic of what had been done and re-write everything from
scratch so that it fit in with the rest of the API ....From there I
just never got around to refining / refactoring things to work more
appropriately...So there...Now I can feel a little less guilty about
things now. The cat is out of the bag. :/

On 2/10/07, Jesse Kuhnert <[hidden email]> wrote:

> I know he is. Time zone differences may delay responses but I'm sure
> they will be given.
>
> Even more importantly, as long as the code changes mesh well with the
> rest of the surefire API I'm sure it will be more well received. Ie
> typical honoring of existing logic && making appropriate overall
> refactorings to the whole api if they are needed instead of making one
> off sort of TestNG specific only kind of logic. (which is more or less
> what I originally did ) ...Just the typical un-fun keeping the whole
> codebase consistent stuff...
>
> On 2/10/07, Alexandru Popescu <[hidden email]> wrote:
> >
> > On 2/10/07, Jesse Kuhnert <[hidden email]> wrote:
> > >
> > > Without knowing for sure how the maven devs feel, I would say that
> > > moving forward with option 1 sounds like a perfectly reasonable
> > > approach. I don't think anyone feels all decisions they make/made are
> > > infallible in some way..
> > >
> > > It didn't work out, move on and do it the right way so the maven2
> > > TestNG users can start getting some sane stability.
> > >
> >
> > I hope Brett Porter is watching the thread and comment on it. If not I
> > will ping him about it.
> >
> > ./alex
> > --
> > .w( the_mindstorm )p.
> >   TestNG co-founder
> > EclipseTestNG Creator
> >
> > > On 2/10/07, Alexandru Popescu <[hidden email]> wrote:
> > > >
> > > > On 2/10/07, Hani Suleiman <[hidden email]> wrote:
> > > > >
> > > > > Is it fair to say that as of right now, the maven/testng integration
> > > > > isn't really ready for most people to use, due to architectural
> > > > > issues with the surefire plugin and a TestNG API that has changed
> > > > > from when the integration was first done?
> > > > >
> > > >
> > > > Without any intentions to sound harsh, I would say that Maven/TestNG
> > > > integration problems are mainly due to the fact that the
> > > > Maven-Surefire code tried to bend TestNG to some different/unsupported
> > > > usage. And I believe that there are only 2 possible directions:
> > > >
> > > > 1/ have the surefire-testng code use TestNG correctly; this way we
> > > > will be able to guarantee the API and the behavior
> > > >
> > > > 2/ have the surefire-testng code continue the direction is started;
> > > > unfortunately I cannot say I would support this direction.
> > > >
> > > > ./alex
> > > > --
> > > > .w( the_mindstorm )p.
> > > >   TestNG co-founder
> > > > EclipseTestNG Creator
> > > >
> > > >
> > > > > On Feb 10, 2007, at 2:24 PM, Jesse Kuhnert wrote:
> > > > >
> > > > > >
> > > > > > This is good news at least. .
> > > > > >
> > > > > > Wrt the junit -like behavior I remember being told something generally
> > > > > > along the lines of "this was done to maintain consistent behavior with
> > > > > > the rest of the surefire API from a user perspective". Maybe this idea
> > > > > > could be re-evaluated given that:
> > > > > >
> > > > > > -) All TestNG related tools (eclipse / ant / idea / etc ) more or less
> > > > > > go through the same basic configuration steps, so trying to change how
> > > > > > that is done is likely going to be an uphill battle with sometimes
> > > > > > b0rken results. .
> > > > > >
> > > > > > -) TestNG seems to be doing a good job documenting how it functions
> > > > > > already, so trying to cater to junit users in the area of
> > > > > > configuration may not provide as much benefit as originally thought..
> > > > > >
> > > > > > -) As long as the plugin does "something" reasonable with little or no
> > > > > > configuration (ie hopefully finding / running a few tests) that would
> > > > > > hopefully be enough to make both sides happy as far as default
> > > > > > functionality goes. ..
> > > > > >
> > > > > > p.s. The maven2 surefire plugin should also be processing tests based
> > > > > > on simple annotation groups excluding suites I hope? (this is how I
> > > > > > often use it at least, ie I have no testng.xml suite definition file )
> > > > > >
> > > > > > On 2/10/07, Alexandru Popescu <[hidden email]>
> > > > > > wrote:
> > > > > >>
> > > > > >> I have had just a bit of time to scan through the existing code,
> > > > > >> and I
> > > > > >> haven't provided any feedback yet because I think the original
> > > > > >> authors
> > > > > >> took a route that was not so easy to grasp for me (dissecting suites
> > > > > >> into tests, etc.). Most of the compatibility problems in the past
> > > > > >> were
> > > > > >> caused by changes in the TestNG internal API, but this was used
> > > > > >> because the original author has tried to "bend" TestNG to work in the
> > > > > >> way JUnit tests are working (and IMO this is not the case).
> > > > > >>
> > > > > >> So, currently Maven2 TestNG code takes a suite and is able to
> > > > > >> disassemble it into <test>s which become runnable pieces. This is not
> > > > > >> correct at least from 2 reasons:
> > > > > >> - this leads to loosing the @Before/@AfterSuite functionality
> > > > > >> - (due to some implementation detail) the names of tests inside a
> > > > > >> suite definition must be unique (limitation that doesn't exist in
> > > > > >> TestNG)
> > > > > >>
> > > > > >> Instead I would suggest that the Maven2 code to behave the way TestNG
> > > > > >> behaves and not introduce new semantics. So, Maven2, should take
> > > > > >> multiple suite, and probably allow separation based on these.
> > > > > >>
> > > > > >> There are more details I have found in the code, but I guess the
> > > > > >> above
> > > > > >> are radical enough to guarantee a discussion :-).
> > > > > >>
> > > > > >> BR,
> > > > > >>
> > > > > >> ./alex
> > > > > >> --
> > > > > >> .w( the_mindstorm )p.
> > > > > >>   TestNG co-founder
> > > > > >> EclipseTestNG Creator
> > > > > > <snipped>
> > > > > >
> > > > > >
> > > > > > --
> > > > > > Jesse Kuhnert
> > > > > > Tapestry/Dojo team member/developer
> > > > > >
> > > > > > Open source based consulting work centered around
> > > > > > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> > > > > >
> > > > > > >
> > > > >
> > > > >
> > > > > >
> > > > >
> > > >
> > > > >
> > > >
> > >
> > >
> > > --
> > > Jesse Kuhnert
> > > Tapestry/Dojo team member/developer
> > >
> > > Open source based consulting work centered around
> > > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> > >
> > > >
> > >
> >
> > > >
> >
>
>
> --
> Jesse Kuhnert
> Tapestry/Dojo team member/developer
>
> Open source based consulting work centered around
> dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
>


--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Jessek

I just hope that people don't expect surefire to start writing their
unit tests for them by the time this stuff comes out .... (whatever
this stuff might be)

--
Jesse Kuhnert
Tapestry/Dojo team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Alexandru Popescu ☀

I am wondering if Brett has any comments on this thread. I tried to
ping him about it, but with no results.

./alex
--
.w( the_mindstorm )p.
  TestNG co-founder
EclipseTestNG Creator


On 2/12/07, Jesse Kuhnert <[hidden email]> wrote:

>
> I just hope that people don't expect surefire to start writing their
> unit tests for them by the time this stuff comes out .... (whatever
> this stuff might be)
>
> --
> Jesse Kuhnert
> Tapestry/Dojo team member/developer
>
> Open source based consulting work centered around
> dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

maven2

Jessek

It's sometimes hard to find people in the right idle state. I only
talked to him for extremely brief chats when going through the first
round of the plugin.

I do remember someone else's name being the most prominent / visible
author tag on most of the source as one calling himself "Jason Van
Zyl". He also talked about surefire sometimes on the maven dev list
(before I unsubscribed - only because my brain is full and I need to
cut back on information). Give him a try.

On Feb 19, 7:55 am, "Alexandru Popescu"
<[hidden email]> wrote:

> I am wondering if Brett has any comments on this thread. I tried to
> ping him about it, but with no results.
>
> ./alex
> --
> .w( the_mindstorm )p.
>   TestNG co-founder
> EclipseTestNG Creator
>
> On 2/12/07, Jesse Kuhnert <[hidden email]> wrote:
>
>
>
> > I just hope that people don't expect surefire to start writing their
> > unit tests for them by the time this stuff comes out .... (whatever
> > this stuff might be)
>
> > --
> > Jesse Kuhnert
> > Tapestry/Dojo team member/developer
>
> > Open source based consulting work centered around
> > dojo/tapestry/tacos/hivemind.http://blog.opencomponentry.com


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: maven2

Alexandru Popescu ☀

On 2/20/07, [hidden email] <[hidden email]> wrote:

>
> It's sometimes hard to find people in the right idle state. I only
> talked to him for extremely brief chats when going through the first
> round of the plugin.
>
> I do remember someone else's name being the most prominent / visible
> author tag on most of the source as one calling himself "Jason Van
> Zyl". He also talked about surefire sometimes on the maven dev list
> (before I unsubscribed - only because my brain is full and I need to
> cut back on information). Give him a try.
>

Yep, I know Jason is one of the core Maven2 guys. Met him in person
and chatted a couple of times, but at this moment I don't seem to find
his contact info around. If somebody can help me out then I will try
to contact him directly and chat about the changes I am suggesting.

./alex
--
.w( the_mindstorm )p.
  TestNG co-founder
EclipseTestNG Creator

> On Feb 19, 7:55 am, "Alexandru Popescu"
> <[hidden email]> wrote:
> > I am wondering if Brett has any comments on this thread. I tried to
> > ping him about it, but with no results.
> >
> > ./alex
> > --
> > .w( the_mindstorm )p.
> >   TestNG co-founder
> > EclipseTestNG Creator
> >
> > On 2/12/07, Jesse Kuhnert <[hidden email]> wrote:
> >
> >
> >
> > > I just hope that people don't expect surefire to start writing their
> > > unit tests for them by the time this stuff comes out .... (whatever
> > > this stuff might be)
> >
> > > --
> > > Jesse Kuhnert
> > > Tapestry/Dojo team member/developer
> >
> > > Open source based consulting work centered around
> > > dojo/tapestry/tacos/hivemind.http://blog.opencomponentry.com
>
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: maven2

Mark Derricutt
He's posting to the maven list with [hidden email].

On 2/20/07, Alexandru Popescu <[hidden email]> wrote:
Yep, I know Jason is one of the core Maven2 guys. Met him in person
and chatted a couple of times, but at this moment I don't seem to find
his contact info around. If somebody can help me out then I will try
to contact him directly and chat about the changes I am suggesting.



--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Brett Porter
In reply to this post by Alexandru Popescu ☀

Sorry for the late reply.

I'll go into some detail (much for my own benefit - concrete stuff at the end).

I agree that it'd be wise to review anywhere that limitations are
imposed on TestNG. It's not meant to be that heavy - the framework is
primarily about making behaviour reasonably consistent (while still
passing through all options), supplying sensible defaults, funneling
output, and lining up the classloaders with the Maven dependencies. So
I think this can be fixed.

Unfortunately, it's buggy, and has gotten worse because of it's use of
internal APIs to try and figure out what classes are runnable as
TestNG tests. I think the first thing to do is fix those bugs.

I think it might be important to define the terminology used as it
seems to be one of the main factors of confusion.

surefire inherited it from junit 3. There's:
- test method = individual test
- test set (test in the plugin, testcase, battery) = class (1..N methods)
- suite = group of test methods (1..N methods and/or classes)

How does this compare to TestNG? As I understand it, they are:
- test method
- test class (1..N methods)
- test (1..N classes and/or methods via includes/excludes)
- suite (1..N tests)

So, they seem to line up if I do this:
- keep current definitions
- rename 'test set' to 'test'
- no need to reference test class directly

One thing I need to check is the directory test set. By default, it
includes **/*.java, and tries to figure out if each class is a testng
class, junit class, or pojo it can run as a test. It does this by
checking if there are any annotations, and this seems to be the
internal API that breaks. We either need to define a public API to use
here, or just send everything to testNG and let it sort out what to
run. The latter seems favourable, esp. since testNG can run junit3
tests.

Next steps:
1) line up definitions
2) check testng.xml properly maps to them
3) change directory based test set to better match testng by default
4) fix outstanding bugs
5) check for internal limitations such as those pointed out earlier
(can duplicate test methods in a test - just need to qualify them if
they are tracked, confirm we can support @Before/AfterSuite).

I've got some solid time set aside for this on a plane next week, so
hopefully I'll make some decent progress on at least fixing the bugs.
Anything we could hash out this week that would facilitate it would be
great.

Also, apologies for not doing this in January like I promised. The
swimming pool was more inviting  at the time :)

Cheers,
Brett

On 19/02/07, Alexandru Popescu <[hidden email]> wrote:

>
> I am wondering if Brett has any comments on this thread. I tried to
> ping him about it, but with no results.
>
> ./alex
> --
> .w( the_mindstorm )p.
>   TestNG co-founder
> EclipseTestNG Creator
>
>
> On 2/12/07, Jesse Kuhnert <[hidden email]> wrote:
> >
> > I just hope that people don't expect surefire to start writing their
> > unit tests for them by the time this stuff comes out .... (whatever
> > this stuff might be)
> >
> > --
> > Jesse Kuhnert
> > Tapestry/Dojo team member/developer
> >
> > Open source based consulting work centered around
> > dojo/tapestry/tacos/hivemind. http://blog.opencomponentry.com
> >
> > >
> >
>
> >
>


--
Apache Maven - http://maven.apache.org
"Better Builds with Maven" book - http://library.mergere.com/

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Hani Suleiman

Hi Brett, thanks for picking this up!

On Feb 20, 2007, at 11:16 AM, Brett Porter wrote:

> surefire inherited it from junit 3. There's:
> - test method = individual test
> - test set (test in the plugin, testcase, battery) = class (1..N  
> methods)
> - suite = group of test methods (1..N methods and/or classes)
>
> How does this compare to TestNG? As I understand it, they are:
> - test method
> - test class (1..N methods)
> - test (1..N classes and/or methods via includes/excludes)
> - suite (1..N tests)
>
How about group support?

> One thing I need to check is the directory test set. By default, it
> includes **/*.java, and tries to figure out if each class is a testng
> class, junit class, or pojo it can run as a test. It does this by
> checking if there are any annotations, and this seems to be the
> internal API that breaks. We either need to define a public API to use
> here, or just send everything to testNG and let it sort out what to
> run. The latter seems favourable, esp. since testNG can run junit3
> tests.
>
Definitely, just pass everything on.



--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply | Threaded
Open this post in threaded view
|

Re: Maven2 Status... again...

Brett Porter

On 20/02/07, Hani Suleiman <[hidden email]> wrote:
> > How does this compare to TestNG? As I understand it, they are:
> > - test method
> > - test class (1..N methods)
> > - test (1..N classes and/or methods via includes/excludes)
> > - suite (1..N tests)
> >
> How about group support?

Good point, that somehow didn't make it from my head to keyboard a
moment ago. Not sure on the best way to handle these - they work now
(to the extent everything else does), by being passed straight through
to testNG. They actually got into the surefire API to facilitate this
- so they should be listed here as part of the standard definitions.

As I understand it, they are a filter (or nested filters) that applies
to a test or tests to execute a pre-defined subset of them - is this
correct?

Can definitely make that part of these definitions, and junit3 can
just ignore it.

Thanks for the reminder!

- Brett

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "testng-dev" group.
To post to this group, send email to [hidden email]
To unsubscribe from this group, send email to [hidden email]
For more options, visit this group at http://groups.google.com/group/testng-dev?hl=en
-~----------~----~----~----~------~----~------~--~---

12