ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tim Vernum <Tim.Ver...@macquarie.com.au>
Subject RE: [Proposal] Sandbox for Wayward Ant Tasks
Date Fri, 04 May 2001 13:52:17 GMT
(the mail actually comes from the commons list, but Peter's
 comment seems the most appropriate one to reply to)

From: Peter Donald [mailto:donaldp@apache.org]

> At 03:14  2/5/01 -0700, Scott Sanders wrote:
> >I will try and initiate that over on Ant-dev.  My intention was to 
> >finish the functionality to enable the xpath task to make it 
> into the 
> >distribution, so an ant-sandox is appropriate.
> 
> But remember that the main reason they are not in the ant project is
> because the committers do not want them there and they think 
> it would be a
> nightmare to support and pain for our users (especially in transition
> between 1.x and 2.0). So realize that there may be a reason for it not
> being done despite everyone thinking it would be a good idea 
> for ant2 ;)

However, it is quite common for the ant commiter to respond to a request
with 
	"It is possible to write a task to do that, but it won't be part
       of ant"

An AntOn/AntCallOn/Foreach task is a common example, and there are others.

In the past month I think I have written 3 tasks, none of which I expect
(or necessarily want) to see form part of the core/optional tasks, but 
which clearly support a defined use case from an ant user.

I think it is quite reasonable to tell users, "We don't like that idea,
but could implement a task for it yourself", but when there is no clear
place to share those tasks, it's a bit hard.

In the last 3 days, I've replied to at least 4 people's requests, saying
"I wrote such a task - search through ant-dev". I don't think that's 
really a good situation.

I have tasks that probably shouldn't be part of the support ant environment,
but which other people want to use.
Should I take these to sourceforge? 
Is that really the solution that people want?

The commiters probably see less need for an "ant-contrib" module, since they
can commit any task they think is useful, but it should be clear to all, that
there are useful tasks that are not being accepted as part of ant.

Even if these tasks don't fit the direct goals of ant, that doesn't mean they
should be lost. 

Other than the adminstration difficulties of managing a new cvs module, what
is the complaint against having a set of contrib tasks? It certainly seems
better than the alternative.



Mime
View raw message