ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Frank_Starsi...@themoneystore.com
Subject Re: How to convince our IT manager to allow ANT?
Date Tue, 11 Apr 2000 19:08:10 GMT


We used Make on a huge Java project and I thought it was a complete pain in the
ass.  Since not everyone was knowledgeable about Make itself, the developers
were
constantly maintaing their Makefiles instead of their Java code... sniffing
around for
extraneous tabs or spaces, making sure all their files were accounted for.. and
in
the right order, getting their environment variables correct, etc..  This was a
horrible waste of time.

Here are some things that I feel make Ant the right choice for Java...


1.  Makefiles need to be constantly maintained when new Java source files
are created and distributed to all other users.  On the other hand, Ant does not
necessarily require any modifications at all when files or even packages are
added to a project, and only 1 file (build.xml) needs to be maintained.
One of our developers had to develop the Makefiles
and the strategy to go with it.  With Ant there is clearly a lot less work that
has to take place
getting it running and maintaining it during a project.
I found there to be a huge configuration issue with Make, especially on NT.

2.  It does not matter whether Ant is not standard at this time. It is simple to
implement
and use. If you decide later to use something else, you've wasted literally no
time
at all.

3. the "makefile" for ant (build.xml) is very easy to understand and Makefile
for the Make
system is very difficult to understand.

4. There are lower support costs for the Ant / Make combo because everywhere you
use
Ant you will be saving money... guaranteed.  Besides, Java is not C or C++. Why
not use
the right tool for the job.

5.  yes, javac does have problems with dependancies. It cannot always be used
alone. It also seems
to me that ANT is faster than Javac.

6. A question of my own...  we develop Java on NT and had to install Cygwin to
get a bash shell so we could use
Make. This was also a configuration issue.  Is there another way to do this that
we missed?








david.boeren@bellsouth.net on 04/11/2000 11:17:12 AM

Please respond to ant-dev@jakarta.apache.org

To:   ant-dev@jakarta.apache.org
cc:    (bcc: Frank Starsinic/TMS)

Subject:  How to convince our IT manager to allow ANT?


I've been testing out Ant for doing our java builds instead of make, and I'm
convinced
that it's a superior tool.  I was able to easily do things that would have
taken me a lot
longer to write in make syntax and the build.xml files are much simpler to
read and
maintain.

Yesterday, myself and one of our system architects tried to convince our IT
manager to
officially allow ANT for building java programs, and met with little
success.

The arguments were:

1. Him: Ant doesn't do anything that make can't do.
   Me:  Because make can call any shell command, there's really nothing make
can't do
        if you work at it hard enough.  My point is that Ant does things
more elegantly.

2. Him: Since Ant would only be used for java builds, we'd need to support 2
tools, Ant
        for java and make for C++.  Therefore, higher support costs.
   Me:  Ant is simpler than make, so it will not have as many problems
requiring a
        build expert as make will.  Also, it can be optional for java, so
not every
        developer needs to be trained to know Ant.

3. Him: I thought javac already automatically detected dependencies and
compiled all the
        necessary files, which means your makefile would generally be
trivial and not
        require a sophisticated tool.
   Me:  First, I've heard of problems with javac with big directories of
files being
        compiled.  Second, I don't think javac can autocompile needed
classes which are
        in different packages (which we use extensively).

4. Him: Make is standard, Ant is not, and may never become a widely used
too.
   Me:  I expect some sort of XML based tool to become a widely used
replacement for
        make, although you can't predict whether Ant will or won't.
Besides, it
        provides value in the short run and increases developer
productivity.  Finally,
        so what if it's not a standard?  As long as it's useful to us, it
doesn't
        matter if other companies are all using it.

So, the question is:  "How do I convince a reluctant manager to see the
value of Ant?"
I am planning to research javac's automatic compiling ability, but I don't
put much
faith in it.  I just didn't have hard proof at the meeting to show.  I can't
quantify
support costs, so I guess I'm going to have to rely on showing him how it
can increase
productivity.  We're going to try to convince the head of configuration
management
probably tomorrow, and then go back to our manager if he will join our side.

Any suggestions?








Mime
View raw message