ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Donald <dona...@apache.org>
Subject RE: Content in tags
Date Fri, 08 Dec 2000 23:32:34 GMT
At 12:43  8/12/00 -0800, Jose  Alberto Fernandez wrote:
>> From: Stefan Bodewig [mailto:bodewig@apache.org]
>> 
>> 
>> Peter Donald <donaldp@apache.org> wrote:
>> 
>> >>But the classpath child shows an area where allowing both, child
>> >>elements and content might be useful - say I wanted to add a
>> >>classpath child to script so that I can point the task to bsf.jar
>> >>and don't need to put it into ANT_HOME/lib or similar.
>> > 
>> > I still think it would be better to seperate them out. Something
>> > like
>> > 
>> > <script language="blah">
>> >   <classpath ref=".." />
>> >   <logic>
>> >    call myScriptTask();
>> >   </logic>
>> > </script>
>> 
>> I know this is the alternative and I could live with it. 
>> 
>> In the <sql> example it means the user would always have to specify
>> <transaction> explicitly, which might make him aware that there is an
>> transaction involved in the first place (which would be a good thing).
>> 
>
>Notice that <transaction> I think means the driver is on autocommit=false
>(actually I am not sure, but that was the intent). So having or not
>having the transaction element makes a difference.
>
>Second, with the case of <classpath> in point. What you suggest will
>mean that if we do not add a <classpath> element from the begining
>to any task someone defines, there is no way to add it afterwards 
>without breaking al previous usage because now you need to use
>some other element that wasn't needed before.
>
>That to me is really really bad.
> 
>I would suggest not being too anal (;-)) about this kind of multiple text
>and let the task decide whether:
>
>addText() means replace, contatenate, or fail is already set.
>
>We shouldn't be in the task meaning bussiness.

Sure we should. We are already in "thought police" mode as a number of
people have pointed out (ie with regards to if/then/else/switch tasks).
Block bad design decisions is a GOOD thing. I am not saying we do it for
ANt1.3 but definetly something to look at for 2.0

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               |
*-----------------------------------------------------*


Mime
View raw message