ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jose Alberto Fernandez" <j_a_fernan...@yahoo.com>
Subject Re: Update to Sandbox/antlib
Date Thu, 14 Feb 2002 16:42:50 GMT
From: "Peter Donald" <peter@apache.org>

> On Wed, 13 Feb 2002 20:14, Jose Alberto Fernandez wrote:
> > > On Wed, 13 Feb 2002 03:07, Jose Alberto Fernandez wrote:
> > > > Proxys need to be there also (optionally) so that we
> > > > can do what TaskAdapter does AND more importantly, try
> > > > to solve the type inconsistency issue between
> > > > TaskContainers and Datatypes that now is solved by
> > > > accident.
> > >
> > > Blech - dont like adapters at all. I guess we can chalk this up to ant1.x
> > > legacy but it still feels ungainly to me :( I would prefer we just ignore
> > > the DataTypes inconsistency and just try not to introduce anymore
> > > inconsistencies.
> >
> > We still have TaskAdapter which converts beans into Tasks. So the concept
> > is still there, and given the rules of ANT1 is there to stay :(
> 
> I dont see TaskAdapter in that light. I see TaskAdaptor as more an 
> alternative strategy for defining tasks. Like that which has been discussed 
> with regards to scripting recently and what was discussed ages ago with 
> respect to templating etc. 
> 

Given that in my code I already have a table of factories instead of a table of classes
and that for adaptors the table already contains the adapted factory, I do not 
really disagree with you on principle. Whether each new task (or something else)
that is added needs to specify its adaptor, or rather the role predifines the adaptors available
is probably a matter of taste. To me the latter puts less burden on the user and
it is more easily backward compatible with ANT1.

Remember that users can register things on their own (even using put/get directly on the table)
and they are not going to be specifying an adaptor. So I am not sure how to
otherwise accomodate what you want (without introducing special cases all over, 
which is exactly what I am trying to erradicate as much as possible).

If you do not want adapters, then you do not need to declare them for a Role.

> 
> One thing you may want to do is have a look at at is JDiff at
> 
> http://sourceforge.net/projects/jdiff
> 
> It allows you to look at the differences between two APIs. It is used by 
> quite a few JSR groups to track evolution of an API. 
> 
> If you could run against your new API and against the old version I am sure 
> that would give people a better sense of how things are changed.
> 
> You could even set up a JUnit test that verified no backwards incompatible 
> changes were made :)
> 
> 

I will keep it in mind. the build.xml is able to produce a distribution installation of
the modified ANT, with that one is able to actually run ANT's QA and that 
is what I am using as the base benchmark. After I am happy with the changes
and JUnit tests, I may ask to try with GUMP and see what happens (now that I saw
that it can be done).

But for the time being my new changes do not work and I am trying to distinguish
between real behavior and unnecessary fixes due to the old datastructures. :-(

I'll let you know when more advance is made.

Jose Alberto

> 
> -- 
> Cheers,
> 
> Pete
> 
> *---------------------------------------------------------*
> | Contrary to popular belief, UNIX is user-friendly. It   |
> | just happens to be selective on who it makes friendship |
> | with.                                                   |
> |                       - Richard Cook                    |
> *---------------------------------------------------------*
> 
> --
> To unsubscribe, e-mail:   <mailto:ant-dev-unsubscribe@jakarta.apache.org>
> For additional commands, e-mail: <mailto:ant-dev-help@jakarta.apache.org>
> 


--
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