geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Prasad Kashyap" <goyathlay.geron...@gmail.com>
Subject Re: [itests] Modify the geronimo-deployment-plugin ?
Date Fri, 04 Aug 2006 16:46:56 GMT
Hi David,

Please see comments inline -

On 8/4/06, David Jencks <david_jencks@yahoo.com> wrote:
>
> On Aug 4, 2006, at 9:00 AM, Prasad Kashyap wrote:
>
> > With the m2migration ready to be merged into trunk, I have resumed
> > work on the itests for Geronimo again.
> >
> > Approx 30% of our code is covered by component level tests that are
> > embedded in each module. These tests are written as junit test cases
> > and run by Maven surefire plugin.
> >
> > The itests will cover system level tests by testing the
> > functionalities that an end-user would use on a fully assembled
> > Geronimo distribution. Therefore to the extent possible, our itests
> > and it's testcases should use the very same external APIs and
> > workflows that a user would use.
> >
> > We have been using the startRemoteServer and stopRemoteServer goals in
> > the geronimo-deployment-plugin (g-d-p)  to start and stop a server. We
> > have always used these "remote" goals and have never used the in-vm
> > goals startServer and stopServer.
> >
> > I propose that we convert the in-vm goals startServer and stopServer
> > to be ant mojos from their existing java mojos. Invoking the ant mojo
> > goals in our itests will ensure that our tests are using the same APIs
> > that a end-user uses. Thus we shall no longer use internal hooks in
> > the code to start and stop the server.
> >
> > Thoughts ? Comments ? Advice ?
>
> I probably don't understand what you are proposing well enough, but I
> don't see how running geronimo in the same vm as the tests will more
> closely approximate the users experience: most people will not be
> running geronimo in the same vm as their client.  Is there some other
> advantage of using only one vm?

Actually I am not saying that we should use only 1 vm. What I am
proposing is that we use ANT to invoke the same CLIs that a user would
invoke. In fact, I'd much rather prefer the start and stop server be
forked off.  I mentioned replacing the existing in-vm goals just
because they were not being used. We may very well leave them there
and add new ANT mojos to the mix.

>
> BTW I am really looking forward to reasonable itests :-)  I wrote a
> couple of tiny sample apps to check if I'd fixed some bugs recently
> and hope these can turn into itests.

Oh cool. Let's talk about this.

>
> thanks
> david jencks
>
> >
> > Cheers
> > Prasad
>
>

Mime
View raw message