On 5/29/08, Simon Laws <[EMAIL PROTECTED]> wrote: > On Thu, May 29, 2008 at 8:48 AM, Rajini Sivaram < > [EMAIL PROTECTED]> wrote: > > > On 5/28/08, Simon Laws <[EMAIL PROTECTED]> wrote: > > > > > > On Wed, May 28, 2008 at 11:16 AM, Simon Nash <[EMAIL PROTECTED]> wrote: > > > > > > > Graham Charters wrote: > > > > > > > >> I've been wondering whether we should make this an itest rather than > a > > > >> sample. We could keep it as a sample, but it relies on > > > >> maven-dependency-plugin to work out the dependencies required to run > > > >> the sample. Is a sample that only works with maven acceptable (I > > > >> believe the other samples do not) or should I change this to be an > > > >> itest? > > > >> > > > >> We do try hard to make the samples work with ant as well as maven. > > > > There have been cases where samples started out with maven support > > > > only and the ant support was added later. From your description, > > > > it doesn't sound lke this is likely to happen. > > > > > > > > I believe the main purpose of this "sample" is to act as a test for > > > > the Tuscany build rather than a sample for a user to copy and adapt. > > > > If this is correct, I think it should be changed to an itest. > > > > > > > > Simon > > > > > > > > > > > > Regards, Graham. > > > >> > > > >> 2008/5/23 Graham Charters (JIRA) <tuscany-dev@ws.apache.org>: > > > >> > > > >>> [ > > > >>> > > > > > > https://issues.apache.org/jira/browse/TUSCANY-2330?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12599389#action_12599389 > > > ] > > > >>> > > > >>> Graham Charters commented on TUSCANY-2330: > > > >>> ------------------------------------------ > > > >>> > > > >>> Hi Rajini, sorry for taking so long to respond. Please go ahead > and > > > >>> check the code in with your update. Changing it to use Felix is > fine > > > by me. > > > >>> I tested it with both and there was little discernible difference > in > > > >>> performance. > > > >>> > > > >>> Thanks, Graham. > > > >>> > > > >>> Calculator sample running in OSGi > > > >>>> --------------------------------- > > > >>>> > > > >>>> Key: TUSCANY-2330 > > > >>>> URL: > > > https://issues.apache.org/jira/browse/TUSCANY-2330 > > > >>>> Project: Tuscany > > > >>>> Issue Type: Wish > > > >>>> Components: Java SCA Samples > > > >>>> Affects Versions: Java-SCA-Next > > > >>>> Environment: All > > > >>>> Reporter: Graham Charters > > > >>>> Fix For: Java-SCA-Next > > > >>>> > > > >>>> Attachments: calculator-osgi-sample.patch > > > >>>> > > > >>>> Original Estimate: 2h > > > >>>> Remaining Estimate: 2h > > > >>>> > > > >>>> It would help with preserving OSGi support if an OSGi sample were > > run > > > as > > > >>>> a matter of course, rather than only by a small number of > > > developers. This > > > >>>> wish is to add the smallest sample possible based on existing > > Tuscany > > > module > > > >>>> dependencies. > > > >>>> > > > >>> -- > > > >>> This message is automatically generated by JIRA. > > > >>> - > > > >>> You can reply to this email to add a comment to the issue online. > > > >>> > > > >>> > > > >>> > > > >> > > > > > > > As we have a distribution that doesn't fundamentally depend on, and > hence > > > demonstrate, how Tuscany might be deployed in an OSGi environment then > I > > > think that a sample that shows how to do this is appropriate. If this > > means > > > that we have a sample that only runs from maven then it's inconsistent > > with > > > our other samples but I could live with that. > > > > > > I guess the real answer is do you think a user could base an OSGi > > > installation on what they learn by looking at the sample. I haven't > > looked > > > at the sample yet myself. Does this bring host-osgi back to life? Is > this > > > sample going to be reworked in the short term as the code is moved > > around? > > > If yes then that would be a justification for keeping it out of > samples. > > > > > > In its current form, the "sample" is too complicated - but it can be > > simplified quite easily to enable it to be used as both a sample and a > > test. > > > > If this is going to be an itest, I would really like it to reuse code > from > > itest/osgi-tuscany rather than create a new copy of the code, requiring > > maintenance of two copies. As an itest, this subset should only add maven > > scripts to create a new set of dependencies. All the code can be used > > straight out of itest/osgi-tuscany rather than through a copy. Since this > > code is likely to change a lot as we tackle versioning etc., and since > the > > calculator subset doesn't really add any new code, it would be much > easier > > to maintain a single copy of the code rather than two (even though both > are > > identical at the moment). IMO, it only makes sense to use a separate copy > > if > > the code is expected to diverge. > > > > > > Regards > > > > > > Simon > > > > > > > > > > > -- > > Thank you... > > > > Regards, > > > > Rajini > > > > Hi Rajini > > Why is the code that is common to the itest and sample in each of these > rather than in a module?
There are two sets of classes that are shared across itest/osgi-tuscany and the calculator subset. 1. OSGi installer - I did think of converting this into a module, but at the moment, this is still experimental code, and we haven't really decided on what the final solution for a bundle-ized Tuscany will be. So itest/osgi-tuscany felt like a better place for it. 2. OSGi test harness - this runs samples (from the samples directory) in an OSGi runtime. This is purely for testing, and doesn't really belong in a "module". If the calculator subset is turned into a sample, it should not really use the harness since it is too complicated. If it is turned into an itest, it should be able to reuse the code directly from itest/osgi-tuscany. Simon > -- Thank you... Regards, Rajini
