ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Erik Hatcher" <>
Subject Re: Ant <project> extension, templatized build files
Date Thu, 02 May 2002 02:04:37 GMT
I know you're very interested!  :)  I wasn't really referring to you, of
course.  The point is that in the Jakarta "culture", we all work on what we
want to work on and our actions speak louder than words - if you have an
"itch to scratch", then you scratch it!

Now that we have the pluggable ProjectHelper, do you see this as an area
where this sort of thing can be accomplished?

Certainly no one can stop you from the preprocessing possibilities such as
XSLT and Velocity - those are all above Ant's core - so you can't blame
veto's from stopping that from happening.  :)

I very much agree that Ant is not providing enough for a complete build
environment.  The fact that others are filling these gaps is a good sign,
especially that there is some competition going too. I just hope those folks
keep in touch with the ant-dev community so that we are aware of any hacks
and workarounds that they are doing because of Ant architectural
constraints. Their work (or pieces thereof) can be leveraged to bring into
the core as their work matures, it seems.

There are other projects layered on top of Ant that folks could claim show
that Ant isn't doing enough as well: CruiseControl, AntHill, Gump, Vindico,
and I'm sure several others. Under the hood of all of them is Ant though.
We just need to be sure to be sensitive and cooperative to these other
projects and make sure we play nicely with them.


p.s. Did you say mutable properties?!   -1    :)))  (just kidding!)

----- Original Message -----
From: "Peter Donald" <>
To: "Ant Developers List" <>; "Erik Hatcher"
Cc: <>
Sent: Wednesday, May 01, 2002 8:46 PM
Subject: Re: Ant <project> extension, templatized build files

On Thu, 2 May 2002 07:04, Erik Hatcher wrote:
> I think its an unfair generalization to say that ant-dev is "unwilling" to
> deal with it.

Historically it is fairly accurate. Go back to december-january archives in
2000-2001, then fastforward a few more months (where it was repeated again)
and then fastforward a few months again ...

> Perhaps "uninterested" is more accurate.

God no. I am VERY VERY VERY interested and have wanted this for ages (about
years). The way I would implement it has changed but not the desire to
implement it ;)

If this had been implemented then we would not even have this flame war on
general. (ie maven vs centipede/gump/forrest). The only reason these
have come into existance is because ant is not providing all the necessary
pieces in a build environment. It is too much work to maintain complex build
files when you have oodles of very similar products and build processes.

> Is there some way
> in which we committers have made it more difficult to build the templating
> pieces on top?

vetoing it? Maybe not you but previous committers have done so and many a
flame war have focused on this issue. The required infrastructure has also
been vetoed a number of times (think mutable properties, scoping, etc).

> I've never seen anyone post any design or patches for integrating any kind
> of templating into Ant's core.

You haven't been round long enough. I have made proposals on a bunch of
different occasions. Everything from making the language dynamic evaluated
and constructed (effectively creating functions in ant) to preprocessing
xslt, velocity) to inline expansion. The only one I haven't proposed for
ant1.x is inline expansion as I guess I learnt my lesson when I proposed the
two previous mechanisms ;)

The recent addition of DynamicTasks makes it possible to bypass many of the
limitations but others still remain.


Peter Donald

To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message