Mailing-List: contact ant-dev-help@jakarta.apache.org; run by ezmlm
User-Agent: Microsoft-Outlook-Express-Macintosh-Edition/5.02.2022
Date: Fri, 23 Jun 2000 04:49:02 -0700
Subject: Re: [RFE] Richer Task Specification
From: James Duncan Davidson <james.davidson@eng.sun.com>
To: <ant-dev@jakarta.apache.org>
Message-ID: <B5789CBE.5534%james.davidson@eng.sun.com>
In-Reply-To: <85256907.003CD475.00@d54mta04.raleigh.ibm.com>
Mime-version: 1.0
Content-type: text/plain; charset="US-ASCII"
Content-transfer-encoding: 7bit

on 2000/06/23 04:04, rubys@us.ibm.com at rubys@us.ibm.com wrote:

> Would the target have to preceed the taskdef in the XML?  Or is he proposal
> to revert back to DOM and not instantiate tasks before they are used?  Are
> there limitations on what tasks may be used in tasks specified as targets
> of a taskdef?  Or am I missing something?

DOM, no. Not necessarily. I think JDOM would work just fine, as would some
private tree built up from a SAX source (or even a little mini parser that
pretty much only understood '<' and '>' and the relatively simple semantics
of the build file so that we could slice our pre-build dependancy on any
particular parser.. but I'm only slightly serious about that and mostly
joking.. :)

But yes, the more we go around, the more that I think that instantiation and
initialization of tasks is definitly a deferred thing.

> My feeling is that while this may be "purer" in some abstract sense, it
> will be harder to explain.  Given a choice, I would prefer something that
> is easy to understand.

I'll go for pure as long as it can be explained in a parapgraph of good
docs. I'm hesitant to go for expediency. It's a fine line obviously. :)

.duncan