Mailing-List: contact ant-dev-help@jakarta.apache.org; run by ezmlm
Precedence: bulk
Reply-To: ant-dev@jakarta.apache.org
To: ant-dev@jakarta.apache.org
Subject: [DISC] ${} expansion for data types
From: Stefan Bodewig <bodewig@apache.org>
Date: 29 May 2001 10:42:19 +0200
Message-ID: <m33d9ov95g.fsf@bodewig.bost.de>
Lines: 25
User-Agent: Gnus/5.0807 (Gnus v5.8.7) XEmacs/21.1 (Cuyahoga Valley)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

We've agreed to unify the namespace of properties and all other
data types - now what are we going to do with ${} expansions, that are
expected to yield a String result.

IMHO for Ant1 like properties it should simply be the String itself.

Likewise, there are some data types where calling the toString method
of the corresponding Object may be enough (like Path, which already
takes the OS into account).  For some I don't think a ${} expansion is
useful at all, but my fantasy is probably too limited 8-).

For all the non-obvious cases we've agreed on pluggable converters (at
least that is my understanding of it 8-).  So Ant would try to lookup
a converter for a given data-type and fall back to using the toString
method if it cannot find one, right?

There are cases where the result a user wants from a ${} expansion
depends on the context.  If you take FileSet for example, people have
asked for lists of absolute or relative paths, separated by a host of
different delimiters.

Does that mean, we want/have to make the converters a hierarchical
aspect?

Stefan