Return-Path: Mailing-List: contact ant-dev-help@jakarta.apache.org; run by ezmlm Delivered-To: mailing list ant-dev@jakarta.apache.org Received: (qmail 18168 invoked from network); 7 Dec 2000 11:29:34 -0000 Received: from mta05.mail.au.uu.net (HELO mta05.mail.mel.aone.net.au) (203.2.192.85) by locus.apache.org with SMTP; 7 Dec 2000 11:29:34 -0000 Received: from cognetnt ([63.34.196.179]) by mta05.mail.mel.aone.net.au with SMTP id <20001207112930.PFLY19572.mta05.mail.mel.aone.net.au@cognetnt> for ; Thu, 7 Dec 2000 22:29:30 +1100 From: "Conor MacNeill" To: Subject: RE: Expansion in includes Date: Thu, 7 Dec 2000 22:27:46 +1100 Message-ID: <006001c06040$b91c0a90$80dc1fcb@cognet.com.au> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook 8.5, Build 4.71.2173.0 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 Importance: Normal In-reply-to: X-Spam-Rating: locus.apache.org 1.6.2 0/1000/N Duncan, > -----Original Message----- > From: James Duncan Davidson [mailto:duncan@x180.net] > Sent: Thursday, 7 December 2000 19:37 > To: ant-dev@jakarta.apache.org > Subject: Re: Expansion in includes > > > Same thing exists with System.properties. "fred" means very > different things > when it is the result of "user.dir" or "home.dir" or > "user.name". Context > matters. I really don't see this as complexity that I'm > advocating. Adding a > bunch of typing on top of this is complex afaic. > System.properties is limited to strings by its nature. You can't store more complex types so, of course, you need to use some context to interpret the meaning. Ant is not (anymore) limited to just string types. A number of other types have been introduced (paths, filesets, etc) and it seems illogical to treat the string type in a different way. > Filesets != properties. A property is a simple name=value > pair. Filesets is > a one-to-set mapping of files which may contain regexp patterns. Well, currently a fileset can also be thought of as a name=value pair. It can be created with a name (id) and can be referred to by that name (refid). Sure that value has a more complex type than a simple string but why should we have to manipulate it in ant build files in another way. If we unify the handling of named values, we should make it more logical and simpler. > The same thing can happen in any data file.. A mistyped > property key in a > java.util.Properties list exhibits this behavior -- as does a mistyped > attribute name in a XML file that isn't being validated > against a DTD. This > is not a problem that is unique to Ant, and yet we're fine > with it in those > cases. But we can do better. If we can catch user mistakes, we should. I still feel that implicit declaration of properties is bad. Conor