I agree with that and having recently spent time helping people new to
Tuscany and seen the problems they've had I think it would be much more
helpful to fail. Could we have a "lenient" mode which can be used when
debugging in eclipse? But I think the default should be strict so when you
deploy a Tuscany webapp it fails if there are issues.

   ...ant

On Thu, Mar 27, 2008 at 11:26 PM, haleh mahbod <[EMAIL PROTECTED]> wrote:

> As a user, I would rely on a composition to behave in certain way.
>  Wouldn't
> bringing up a composite for which some of the components don't work the
> way
> they should, invalidate my assumption and lead to unexpected behavior at
> runtime?
>
> On 3/27/08, Jean-Sebastien Delfino <[EMAIL PROTECTED]> wrote:
> >
> > Vamsavardhana Reddy wrote:
> > > In that case do we need to make sure that the spec (the next version
> of
> > it)
> > > says the same by getting the errata[1] corrected?
> > >
> > > [1]
> > >
> >
> http://www.osoa.org/display/Main/Errata+for+Java+Annotations+and+APIs+V1.00
> > >
> > > ++Vamsi
> > >
> > > Jean-Sebastien Delfino wrote:
> > >> +1 to report a misplaced annotation as a warning.
> > >>
> > >> Throwing an exception that'll prevent my application to run just
> > because
> > >> a harmless annotation was present (and not considered) somewhere in
> > part
> > >> of my code seems too aggressive to me.
> > >> --
> > >> Jean-Sebastien
> > >>
> >
> > I think that the statement "It is an error to use this annotation on an
> > interface." in [1] is correct. It is a programming error.
> >
> > The only thing I'm saying is that the Tuscany runtime should not prevent
> > a composite to be deployed and started when one of the classes it
> > references contains a programming error.
> >
> > IMO the runtime should warn the user with a 'Hey you've got a
> > programming error in one of your classes, an annotation is mis-placed
> > and we're ignoring it'.
> >
> > On the other hand an Eclipse tool for example should probably report
> > that programming error as a problem with severity=error in its problems
> > view.
> >
> > --
> > Jean-Sebastien
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>

Reply via email to