From: "Peter Donald" <[EMAIL PROTECTED]> > On Mon, 22 Oct 2001 18:01, Jose Alberto Fernandez wrote: > > From: "Peter Donald" <[EMAIL PROTECTED]>
> > > And tell me - how many times have the committers stated that they hate > > > this? or that this is an ugly hack ? > > > > I have not heard anyone. > > I recall myself and Setfan saying so on the public lists. I am not sure about > Connor though I thought he disliked treating targets as methods ??? The only > committers I can remember who have actually supported this sort of structure > was me and Diane. Ironically - I am told it was you who convinced me that > this was a bad idea. > I may need to be refreshed about the context in which I made my argument. I doubt I was suggesting that <antcall> was always bad. I certaintly think that for many of these issues the devil is on the details (and how orthogonal are the ideas behind it ;-)). > > In particular given the fact that the above line > > could have been rewritten as: > > > > <target name="main" depends="dist-lite" > > description="--> creates a minimum distribution in ./dist" > > /> > > > > 8-) > > true - but in other cases the reason for using antcall was to work around > limitations of ant. > Like not wanting to repreat the same set of target 20 times just because they need to be applied with slightly different property settings. Oh but wait, wasn't that the original prehistoric reason having subroutines? > > The point of the matter is that eventhough the above situation may be > > rewritten to avoid an <antcall> there are plenty of sitiations where > > <antcall> is completely valid. And where <antcall> is the more efficient > > and clear way to express what is required on a build. > > And there is plenty of times when perl is the right answer to a question and > java is not. However that does not mean java should change to be more perl > like. > Nor it means that Java should not have good pattern-matching support (i.e., ORO). Jose Alberto
