ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Magesh Umasankar" <umag...@apache.org>
Subject Re: Backwards incompatible change to Javac
Date Fri, 22 Mar 2002 15:31:40 GMT
From: "Stefan Bodewig" <bodewig@apache.org>


> A while back I've changed javac's fork attribute from boolean to
> String - so that one can specify the name of the executable to fork.
> I never thought anybody would be using these method directly, I was
> wrong: <http://jakarta.apache.org/builds/gump/2002-03-21/castor-test.html>
>
> We cannot simply add
>
>     public void setFork(boolean b) {
>         setFork(""+b);
>     }
>
> as IntrospectionHelper then will chose this method over the one taking
> a String argument - which means fork="/my/path/to/javac" would become
> the same as fork="false".  Not good.
>
> The only solution I can see:  change fork back to boolean and add a
> new attribute for the forked executable.  This again is backwards
> incompatible to the change I've made more then four months ago and
> people may have started to depend on it.
>
> Any ideas?
>

+1 for new attribute.  In fact I am used to
setting fork to true or false - backwards incompat
apart, overloading its meaning implicitly might not
send a strong enough message as explicitly adding a
new attribute will.

Just make sure the Castor folks do not
change their call to use String instead to work
around the issue they are currently facing ;-)

> Stefan
>


**********************************************
*  Etc.: A sign to make others believe that  *
*  you know more than you actually do.       *
**********************************************



--
To unsubscribe, e-mail:   <mailto:ant-dev-unsubscribe@jakarta.apache.org>
For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>


Mime
View raw message