On Mon, Apr 11, 2011 at 1:02 PM, Simon Nash <[email protected]> wrote: > ant elder wrote: >> >> For the past 2.x releases we've acknowledged that the samples haven't >> been perfect but have done releases anyway, are you suggesting we >> should stop doing that and get the samples more perfect before any >> more releases? I'd like to do a release now while we continue to work >> on the samples. I also think it would be great if you could actually >> help write some samples in SVN along with all these emails about them. >> > I'm not saying that we should wait for perfection in the samples before > doing more releases. As I've said earlier in this discussion, I think > it's fine to do releases with a small number of samples that meet some > kind of minimum standard and add more samples in future releases. > > This discussion is about what's expected of a sample before it gets > included in a release. It looks like we are converging on an agreed > checklist for that. I hope this will be useful for deciding which > of the samples are ready to be included in a release. > >> One other comment inline below... >> >> On Thu, Apr 7, 2011 at 2:59 PM, Simon Nash <[email protected]> wrote: >>> >>> Thanks for reviewing the checklist. See comments inline below. >>> >>> Simon >>> >>> Florian Moga wrote: >>>> >>>> The checklist is looking good until now. I think that if we want to >>>> enforce a consistent feel over the samples we should agree on build >>>> tool, >>>> dependencies and launcher before considering the checklist final. >>>> >>>> I would say let's stick with Maven for now, Apache Ant adds complexity >>>> and >>>> requires time to find an elegant way of writing a script which we don't >>>> afford now. >>> >>> I think we also need to show people how to build samples using the >>> binary distribution without needing to download additional modules. >>> This is the reason why we have previously provided ant scripts for >>> building the samples. I don't think it's necessary for every sample >>> to have an ant script, but at least some of them should have one. >>> >>>> As for how dependencies are declared, you've got much more experience >>>> with >>>> Tuscany to weight the pros and cons for each approach but I think we >>>> should >>>> use the one we consider best practice and present that to the user. For >>>> me >>>> base+extensions seems to be the way to go as it looks more loosely >>>> coupled >>>> and there's been the big effort of adding that in Beta1. >>>> >>>> Regarding the launcher, I have already expressed my opinion a couple of >>>> times but haven't seen any comments so I guess we're fine with the >>>> shell? >>>> >>> This is probably OK for developers but it isn't suitable for deploying >>> or embedding Tuscany. I think it's important that we include ant scripts >>> for running some of the samples so that people know how to run Tuscany in >>> a production or embedded environment. >>> >> >> I think the Shell is a great way for users to learn about using SCA >> and things like contributions and composites, better IMHO than the Ant >> scripts. It doesn't show how to embed Tuscany but that not its role >> and we can have other samples that show that. >> > +1, the shell is good for people learning SCA, and ant scripts are good > for showing how to embed Tuscany. A mixture of both styles in the samples > is a good approach. > > Simon > >> ...ant >> >> > >
The requirements/checklist looks like a really good idea to me. Re. launching. I suggest we have a separate set of samples that specifically demonstrate the numerous different ways of launching Tuscany. Shell/maven/ant/OSGi... This then removes the need for all other samples to each demonstrate multiple approaches to sample launching. The majority of samples can rely just on shell (or drag and drop in the web app case). There are of course exceptions which we'll just have to address on a case by case basis. Maybe when we introduce the samples for launching Tuscany we document the rationale for choosing each approach and then just follow our own advice. Simon -- Apache Tuscany committer: tuscany.apache.org Co-author of a book about Tuscany and SCA: tuscanyinaction.com
