On Tue, 29 Aug 2006, Jesse Glick <[EMAIL PROTECTED]> wrote:
> Stefan Bodewig wrote:
>> Sometimes we build new features that are useful enough to apply the
>> right away in Ant's own build file.
> 
> Obviously. But it seems to me that the convenience of the many users
> of Ant (sources) outstrips the convenience of the few Ant
> developers.

I don't really expect the number of users that try to develop Ant's
trunk version to outnumber the Ant developers.  Even if it did, you'd
just be asking for classloader problems if you have two different
versions of Ant inside the same Java VM while compiling Ant or running
the Unit tests, aren't you?

> For example, it is impossible to develop the Ant trunk in NetBeans
> using the bundled Ant 1.6.5 unless you write a wrapper script which
> execs bootstrap.sh, which is unpleasant.

Can't comment on that since I have a personal preference for more
arcane development environments myself.  I certainly don't want to
make Ant development inconvenient for anybody, though.

> Similarly you could not use a bundled Ant binary on Linux to run
> ant/build.xml targets (or use Bash completions...).

The first thing I'd do is throw out the bundled binary and install the
last released version myself anyway ;-)

> To Kev: surely most Ant committers work on at least one other
> Ant-based project where improvements can be seen? Or does the Apache
> Foundation pay your rent? :-)

My rent is paid by working on projects that don't use Ant at all 8-)
And where I actually use Ant at work, I wouldn't be allowed to use a
non-released version, so no, the improvements can't be seen until
after the release of a new version.

> Just wondering why an Ant release is expected to be used by many
> thousands of Java programmers for every purpose you can think of,
> yet it's not good enough for Ant itself which is a pretty small
> project.

Ant is certainly good enough for Ant, its just that the next version
of Ant is always better - for many thousands of Java programmers as
well as Ant itself - than its predecessor.

Stefan

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to