ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Rees <>
Subject Re: [DISC] Task API
Date Tue, 10 Apr 2001 03:06:36 GMT
On 02 Apr 2001 11:36:59 +0200, Stefan Bodewig wrote:

>Peter Donald <> wrote:
>> Marker interfaces only go one level deep easily - if we use TOMs
>> then we can do things (as ugly as they may be) like
>> <if test="...">
>>   <while test="...">
>>     <third-layer-task ../>
>>   </while>
>> </if>
>> or absolutely anything really ;)
>Making <while> know its children, but not <if> know its grand children,
>so what? If <if> needs to know, then <while> could expose them.
>Still cannot see the benefit of all this, make my vote a -0 as I don't
>really see a reason (apart from API bloat) against "give tasks access
>to their Task Object Model" either, so I couldn't really justify a

I would think that an extended task API beyond create could be used
for flexibility we need without having to go to a full XML parse by
the task. Something that passes the sub-element name and the suggested
class name (based on however we are reading these mappings in) to the
class along with the helper. The default behavior is imply for the
Task to turn around and tell the helper to do its magic, but
sub-classes could override, perhaps still calling the helper API after
making some changes (like changing the class).


View raw message