ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Donald <>
Subject Re: [RFE] Richer Task Specification
Date Thu, 22 Jun 2000 12:39:22 GMT
At 03:01  22/6/00 -0700, you wrote:
>on 2000/06/20 01:45, Peter Donald at wrote:
>> well if I was a voting member I would go +1 but because I aint I will just
>> say good :P. Thou don't see why the properties/taskdefs don't get treated
>> exactly like tasks (except that maybe they can never be overidden) so are
>> you asking for just a philosophical distinction and safety net so some-one
>> cant add their own task named taskdef ???
>No, a chicken is not a worm is not a mammal. Properties != Taskdefs !=
>Tasks. Sure, you could drive taskdefs from properties (or even derive such
>as taskdef extends property but since this is XML and not Java...), but I
>think that they really should be seperate beasts and treated as seperate
>collections of things.

I understand they are philosophically different and should be treated as
differently by ant-processor but ... They could be implemented as tasks and
executed immediately (ie when you get to them). They would still be treated
differently because you won't be able to have arbitary tasks in main part
of build file only properties/taskdefs/target tasks.  Implementation wise
it is _good_ IMHO(virtually no special case code) and it could be still
kept seperate philosophically.

I still don't understand properties thou and none of the archives seem to
explain it (maybe I just thick :P) so I would love for some one to explain
it as it stands now. I saw references to removeing ${}, making them
entities treating the variables through available and properties as
constants but I still have no idea how it was decided (or if it was) what
they are going to be.



| "Nearly all men can stand adversity, but if you want |
| to test a man's character, give him power."          |
|       -Abraham Lincoln                               |

View raw message