ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matt Benson <gudnabr...@yahoo.com>
Subject Re: Ant "roles"
Date Tue, 19 Apr 2005 15:28:47 GMT

--- Stefan Bodewig <bodewig@apache.org> wrote:
> On Fri, 15 Apr 2005, Peter Reilly
> <peterreilly@apache.org> wrote:
> 
> > I do not like roles
> 
> I'm not sure we are really talking about roles at
> all here.
> 
> What Matt and DD and I and others who've spoken up
> mostly want are two
> things:
> 
> (1) Be able to use <istrue> via add(Condition).  I
> had to extend
>     ConditionBase to make AssertTask accept Ant's
> built-in conditions,
>     this is cumbersome.
> 
> (2) resolve naming conflicts we'd get when we have
> multiple
>     definitions of the same name.
> 
> DDs idea to have antlib individual descriptors for
> conditions,
> selectors and so on that the user can use simple
> works here.  You can
> still use the existing tasks without any namespaces
> and you can
> explicitly load conditions into a namespace if you
> want to (or have to
> to resolve conflicts).  This really would already be
> enough for me.
> 
> What brought the term "roles" was that DD didn't
> have a better term
> for the collections of things that are together in
> one such antlib
> descriptor.

I think we already have 1. and 2. if we want to use
antlibs, and assuming we can place additional
resources where we like.

> 
> And then we'd really like something shorter, sweeter
> than
> 
> <project xmlns:c="myconditions">
>   <typedef
>
resource="org/apache/tools/ant/types/selectors/typedefs.xml"
>            uri="myconditions"/>
> 
>   ...
>   <condition>
>     <c:istrue ...
> 
> Something where loading of the descriptor gets
> triggered by the
> namespace URI, but this is optional, at least for
> me.

If we have consent to add resources, then yes, the
above is optional, but for me only barely so.  Peter's
concerns of cumbersome-ness would be realized IMHO by
forcing such long declarations.  It almost seems
integral that if we are going to essentially bundle
antlibs in the core, then those should be
distinguished by a custom means of access, and that as
terse as possible.

Nag, nag, nag... but I'm very frustrated by this
stalemate.

-Matt

> 
> Stefan
> 
>
---------------------------------------------------------------------
> To unsubscribe, e-mail:
> dev-unsubscribe@ant.apache.org
> For additional commands, e-mail:
> dev-help@ant.apache.org
> 
> 


		
__________________________________ 
Do you Yahoo!? 
Plan great trips with Yahoo! Travel: Now over 17,000 guides!
http://travel.yahoo.com/p-travelguide

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ant.apache.org
For additional commands, e-mail: dev-help@ant.apache.org


Mime
View raw message