ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Donald <dona...@apache.org>
Subject Re: That's what I was gonna say....
Date Tue, 19 Dec 2000 01:41:04 GMT
At 10:27  18/12/00 -0800, Diane Holt wrote:
>I realize I'm only a lowly end-user (with a couple of very minor patches
>submitted) and I don't have any voting rights, but if I -were- able to
>vote, I'd vote for this.

ant-dev is meant to be slaves to end users so ... ;)

>Now that Ant has been getting used in the Real World for awhile, it seems
>to me that the greatest resource available to help in the design of an
>Ant2 would be to gather information from users -- find out how they're
>using it, what they like, what they don't like, what they need to do and
>currently can't, etc.

Right - surveying ant-user is good start. I also have a lot of
semi/non-technical people who use it who give good feedback which I hope I
pass on ;)

>I've been trying to keep up not only with the proposals as they're
>submitted but with the discussions about Ant2 in general and the various
>proposals in particular, and I'm sorry to say that, at this point, I still
>have no clear idea of what any of them will or won't do for me with regard
>to what I will or won't be able to do with an Ant2 that I can't currently
>do with Ant1.n -- and there are definitely things I can't currently do
>that I would very much like to be able to do.

Well not much as a user. Hopefully they will be as painless to migrate as
possible. If you notice a difference as a user (and not a developer) then
we have possibly failed to a small degree (or large one). However Ant2.0
should enable other people to provide more facilties. ie the templating you
ask for below, easy integration with GUIs and/or IDEs, consistency between
all tasks, easy facades (so you could use your own javac implementation to
get around our crappy jikes ;]). However all these features will only be
easy to implement if Ant2.0 is based on the "right" architecture. Given the
"right" architecture then near anything is possible. Thats where we are now. 

Some people don't think that the core should be "compromised" to allow ant
to easily integrate with GUI. Others don't think templating is a good thing
and still others (I think just JDD at this time) think we should revert
everything to simple string manipulation. We have to find a way to satisfy
all parties.

The only additional things that may make your life easier that have so far
been suggested is that -
* you never have to define ANT_HOME unless on win9x
* AntFarms idea of imported projects means you can compartmentize build
files and have smaller more manageable build files.

>I also very much miss having a simple way to include other files (the
>ENTITY thing doesn't really do it for me -- 1) it's not simple, and 2) it
>doesn't resolve property references for the filename).

AntFarms solves this to a small degree while templating does it to a larger
degree.

>The other thing I'm not thrilled about with Ant is the XML build-files --
>they can be rather daunting to people who are used to dealing with, say,
>Jam build-files, which can be as simple as, for example:
>  Javac foo bar baz ;

Well I guess this is a style thing and I can not see it changing ;)

>I'm not saying I -prefer- Make or Jam over Ant -- if I did, I wouldn't
>have just gone through a massive conversion of a Jam-based system to be an
>Ant-based one. There are *definitely* advantages to using Ant for a
>Java-based product, and I fully intend to continue using it -- I would
>just be happier to see a greater focus, at this point, on -what- Ant does
>and doesn't do than on the -internals- of how that's to be accomplished.

It is the internals that decide what Ant does thou ;) Only the right
architecutre can provide the necessary features to do all these things
without unecessary hacks and or massive reinvention of the wheel.

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