[prev in list] [next in list] [prev in thread] [next in thread] 

List:       tuscany-user
Subject:    Re: Tuscany 2.0-Beta1 sample can not run (UNCLASSIFIED)
From:       ant elder <ant.elder () gmail ! com>
Date:       2011-02-02 14:59:28
Message-ID: AANLkTinniez9aFfmZiytNCHuHCpKmePPTMvYLGAkCWHn () mail ! gmail ! com
[Download RAW message or body]

On Wed, Feb 2, 2011 at 1:58 PM, Simon Laws <simonslaws@googlemail.com> wrote:
> On Wed, Feb 2, 2011 at 1:41 PM, Richard Yao <Richard.Yao@teranet.ca> wrote:
> > For "(5) When the sample reorganization happened often relativePath was
> > not
> > updated correctly so now the parent can't be found"
> > 
> > Just curious that do we have some kind of "continuous build" to make
> > sure new change won't break old code?
> > 
> > I am doing a POC for SCA and having worked on both Fabric3 and Tuscany
> > for the past few days. So far, I feel Fabric3 is easier for newbie
> > because their samples can pass the build and be used just as is. Both
> > projects do not have a lot of detailed documents and samples are the
> > documentation itself. So having working samples is very very important
> > because even I want to spend more time to dig but my boss won't allow
> > me.
> > 
> > Just as I said, I am a newbie and sure some my statements could be very
> > wrong.
> > 
> 
> I think you're right Richard. We need to be running the samples from
> the build as we expect the user to run them. We have a place to do
> this (itest/distribution) but not all samples are included in that
> yet. If we drive them from there then they will get run Hudson. At the
> moments the samples only get built as part of the main top down build
> which is not precisely how the user sees them.
> 
> Simon
> 

I think we all agree the samples need work but as some justification
we have been focusing more on the runtime function and leaving the
samples a little till closer to the 2.0 final release. Thats getting
close now and the samples are getting more time spent on them now so
they should get much improved.

The continuous integration testing is interesting, if you look at some
that do already have an integration test eg [1] that still had a
broken parent in the beta1 release. The reason is that integration
test picks up the maven environment from the running build but the
problem will only happen in an environment where a tuscany build
hasn't been run. I guess the test could spawn a second maven process
using an empty repository but that would add a lot of time to an
already lengthy build time. It looks like Maven 3 does better
validation of the parents so maybe just doing the nightly build with
Maven 3 will fix this problem. We could also look at why/if the
samples need a parent, we have some which don't eg [2] so maybe thats
a better approach.

   ...ant

[1] https://svn.apache.org/repos/asf/tuscany/sca-java-2.x/tags/2.0-Beta1-RC3/samples/learning-more/implementation-java/calculator-contribution/pom.xml
 [2] https://svn.apache.org/repos/asf/tuscany/sca-java-2.x/trunk/samples/applications/eightball-demo/eightball/pom.xml



[prev in list] [next in list] [prev in thread] [next in thread]