I think the main problem with <description> is that we are conceptually treating it wrongly. <description> is conceptially a syntactic element of <project> and <target>, and not a regular task or <datatype>. The fact that was implemented as a <datatype> was just a matter of convinience at the time.
If we were to treat <description> as a syntactic element (i.e., ProjectHelper* must apply them but they do not show in the execution tree), then the only drawback is that they will have the same status as <target> or <project> and could not be redefined as something else. I think this would be the correct thing to do. Jose Alberto > -----Original Message----- > From: Costin Manolache [mailto:[EMAIL PROTECTED] > Sent: 12 December 2002 23:13 > To: Ant Developers List > Subject: Re: Delayed element creation implementation > > > I see one failure - descriptor4. > The test case is: > <project> > <description>a </> > <target> > <description>b </> > </> > </> > > And the test expects to see "a b" as description. > > IMO this behavior of description is wrong. I could > buy the description inside target setting the description > for the target, but having it added to the project descr. > is at least strange. > > If this is not the case - I can try a workaround ( like > walking the tree specifically for description elements ). > > Costin > > > Costin Manolache wrote: > > > I'm ready to commit an implementation of the proposed > > "delayed" element creation and top level execution order. > > > > It is based on the embed ProjectHelperImpl2 - I cleaned > > up every other feature ( dynamic properties, component factories ) > > and remove the 'legacy' code that supported the old execution > > model. > > The changes can be applied to ProjectHelperImpl, but I think > > using the new code is better - and it gives us SAX2 and namespaces > > ( not used yet - but available ). > > > > I'll make sure the tests are passing and do a gump run > > on my machine, then commit. As usual, if you see any problem > > ( or if the official gump does ) I'll revert the change > > imediately. > > > > Please review, it's an important change. > > > > Costin > > > -- > To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
