At 07:58 12/1/01 -0500, Sam Ruby wrote:
>One think I very much like about ant is it's near zero learning curve. I
>just got another convert yesterday, and it always is a joy to behold. What
>helps make this possible is the fact that there is almost no hidden
>knowledge about how things work inside required to get started. Much of
>the original implicitness, like the "init" task and the automatic copying
>of non-java files into the target by javac have long since been removed.
yup ;)
>The build.compiler is an exception. It "talks" directly to the inner
>workings of a specific task, without much in the name giving a visual clue
>as to how it works. For example: does it apply to the rmic task? No.
>
>For starters, I would prefer a compiler attribute to the javac task.
agreed.
>Then I would go a step further, and make it default to the value of
>${javac.compiler}, if present. For now, we could have property
>javac.compiler default to the value of build.compiler, and deprecate it.
>
>Taking this a step further, I would make any task's attribute default to
>${<task>.<attribute>}. Now I can control deprecation on compiles, the
>window title on javadoc, the useTimestamp option on get, the jvmarg on
>java, ... all from the command line!
+1
Remember ages ago when I was talking about either CSS or default proeprties
- this was exactly what I meant. When Constructing task objects from proxy
then you would set all explicitly defined attributes for task. The you set
all implictly defined attributes that haven't already been explicitly set.
I suggested this pre-comimtter status but it was -1'ed back then IIRC
because it was too complex ;)
>P.S. Yes, I am the creator of the build.sysclasspath property. So,
>sue me. ;-)
> [I would have prefered to name it project.sysclasspath]
Hey I have probably been mostly responsible for jikes specific properties -
(ie I have done much more worse than you ;-])
Cheers,
Pete
*-----------------------------------------------------*
| "Faced with the choice between changing one's mind, |
| and proving that there is no need to do so - almost |
| everyone gets busy on the proof." |
| - John Kenneth Galbraith |
*-----------------------------------------------------*
