It's a relatively thin wrapper around Javac for defining your dependencies and implementing a standard compile, package, test, deploy lifecycle.

In the bad old days you would just half-implement the same thing in an ad-hoc way.

Various Maven modules were restricted, including Antrun. Because the projects are old and tied to their directory structure of resources, it was more practical to script in Maven rather than follow conventions. Scripting in Maven was quite... challenging.

In this case, the problems arising from Maven also stems from the organization's context.

After having jumped into Rails and shifting from BBEdit to TextMate as my primary editor overnight back in 2005, when TextMate started withering on the vine I became disillusioned that I had put so much effort into pursuing such a short-lived tool.

Reflecting on my history with a UNIX shell going back to the late 80s, I realized that things I had learned 10, 20 years ago from the UNIX world were still relevant today. I committed myself to getting serious about vim because I want to optimize for A) learning many programming languages and B) not using verbose Java-like languages that require IDEs for the all the boilerplate and rote refactoring.

While this kind of toolset will never provide quite the bang for the buck of a contextual IDE in a specific language, it's a phenomenal hedge against all the career risks I face in terms of Ruby becoming irrelevant, the web becoming irrelevant, Apple nerfing OS X, or any other probable sea change. No matter what happens I feel like vim + bash will bring me an immediate level of productivity in any new task I face, even if I start flattening out before I reach the Eclipse or Visual Studio level of wizardry, I don't expect any one thing to last long enough in this industry for such optimizations to pay off.

Recently, I inherited a project that had 10,000s of lines of ant scripts. Resource constrained, I didn't have anyone to maintain that build process. So, I took a few hours and converted those projects to several hundred lines of maven 'scripts'. Now, it works way better than the previous system, because now I can have a person focused on coding features, rather than maintaining process.

Maven was what I knew at the time; and, it fit well enough. Maybe Grails is the right answer, though. Or, something else.

Anyway, ultimately, it comes down to what trade-offs we make and what things we prioritize with our limited knowledge that we have available. And, that is the art of software development.

But he is absolutely correct about the key observation: a basic tool with "plugins" is simply NOT the way to create a build tool. A built tool should allow (when needed) the full flexibility of a Turing complete programming language. Because eventually your project is probably going to need it.

Anyway, you have three levels of builds.

- Complex Perl scripts, bash/unix scripts : these people are insane and generally have degrees in physics and mathematics. They write scripts that are unreadable, can't modify, are cryptic and I imagine that that the authors can't read them. I sometimes see these in old Linux programs.

- Complex Make, Autoconf, bash/unix scripts : A little better than the perl build scripts but still a complex because you have to understand each command used. I am not fan and never was a C/C++ developer. I guess these are still used in the non-Java world. I am looking at you 'OpenJDK/Sun JDK'

- Complex Ant scripts : These aren't bad, easy read, a little bit difficult to maintain over time. I think most developers prefer Maven but don't mind an Ant script if the project is small

- Maven Scripts : I think people use maven because not much else exists. And the problem of compiling code is not that serious as the actual application development.

And then the person that wrote them leaves and we're fucked. If maven doesn't do what you need to do, then chances are you don't need to do it. You may think you need to do it, you probably don't. Maybe what you need is a more powerful deployment system, or something that takes over after your java/scala/whatever is built.

If you're building perl, C, C++, sure, maven isn't going to do it for you. If you're building java, you're sorted.

My favorite is when this sequence happens (I have seen it first hand):

  1. Maven is this external system that we have no control over.
  2. We will write our own system that will be awesome.
  3. Wow, look, every project uses our Build System!
  4. Oh, look, someone on project X changed B.S. and now Projects A-N don't work!
  5. We will lock down B.S. so only the one person who wrote it can change it!
  6. ...
  7. B.S. is this internal system that we have no control over!
  8. ...
  9. Can we add support for merging war files?
  10. No that'll take weeks, and Bob isn't available anyway.

I work in a java shop currently. There's got to be a hundred of us. Every single one uses an IDE.

When I use maven, my IDE just works. eclipse, intellij, netbeans all support maven. Want to put a bunch of files in a common war to share? maven has a merge-war plugin and intellij and eclipse know what it means: they both build projects that copy the files just like maven would do. Better yet, as you edit those files, no matter where they are, both eclipse and intellij update the files in the deploy target and they are available when you refresh your browser. There is a huge amount of investment in making this stuff work. This is convention over configuration.

No tool can do this with Ant (or Ruby) unless it runs the Ant and the Ruby, and then you've side stepped the IDE and reduced them to dumb text editors. With an Ant project, the IDE can't know that when I edit file A, it is transmogrified into file Z. (If it could, it could solve the halting problem). With Maven, the IDE implements convention, and so it know that File A is compiled to File B and then must be copied to location Z because its part of a merged war. Add JRebel to the equation and pretty much any change to the code is immediately runnable. Its dangerously easy to spike away instead of writing tests!

Instead, you want general purpose language that builds the build dag using the build system as a library. This gets you control and customisability for free, and you can still get reuse through code libraries.

Last year replaced a maven shitfest with an old fashioned Makefile which calls javac and manual dependencies. The whole port process took about 2 hours.

Integration and test run time has gone from 7 minutes (to drag all the maven plugins and dependencies down) to just over two minutes.

This saves us a fortune on build agent capacity as well (we use TeamCity).

I've worked on countless Maven builds and have done everything under the sun. Maven's plugin system allows you to ultimately achieve whatever you need out of your build--in the 10% case that you have a need that isn't already covered by the varied plethora of plugins out there. You can even call out to Ant using the antrun plugin.

I've found so much simplicity, reliability, continued payoff in my use of Maven. I've heard complaints but usually from people that refuse to understand how it functions declaratively.

Declarative systems like Maven ask that you learn something first, and then reward you infinitely if you do; imperative systems let you do something immediately without learning a model, but then make you infinitely pay for it.

Tools like Rake and Gradle (and Buildr) are more like thin wrappers on a turing complete language and are able to stretch and bend to fit the problem. Sure, that makes them harder to get started with, but it provides much more runway when the build gets complicated.

I've used Maven in the past on some projects and it starts off very nicely. However, without fail, I end up with some part of the build that doesn't work quite right. Or throws unresolvable errors. Or just fails, some of the time. I guess I (and everyone I've ever worked with) sucks at Maven and just can't make it work. If so, that tells me much more about Maven than my team's abilities.

P.S. I (mostly) blame XML for my Mavin chagrin. Over the years I've grown to hate XML and everything associated with it. It should burn in a fiery pit of lava. But that's just me. :)

I'm not sure if it's due to someone's lack of technical ability, or if they get distracted by other externalities, but we always tend towards something that seems really simple at first but instead incurs a lot of technical and process debt later on.

Brilliant read.

It gives us the language to efficiently reason about and discuss build concepts with colleagues.

This is a big thing, think about the most recent new starter situation you experienced, about how communication of these common build tasks was expressed. Chances are if it was with a more composable tool, the discussions were fairly detailed, the words 'just like' frequently appeared. If it was with a more rigid tool, the conversation would have been in terms that tool commonly defines.

A great example is the present state of browser development t. Google "backbone sucks site:news.ycombinator.com" and you find tons of comments from people who struggle with backbone - a supercomposable library, not a framework by design.

Composabilty pays off huge in the long, but you need experienced leadership to wield it.

There's also a happy medium in things like Play Framework, a functional scala framework but designed almost as a collection of composable libraries so you can lean on Plays opinions for most things but since it is designed to be composable (compare to "pluggable", or "extensible") so you can swap different pieces out if you like. That is my experience, at least.

tr A-Z a-z | tr -cs a-z '\n' | ...

Since you first turn upper-case characters into lower-case ones, you then get to replace only things that aren't in a-z with newlines, rather than things that aren't in A-Za-z. It saves a whopping three characters!

http://www.mail-archive.com/users@maven.apache.org/msg126205...

At least the people behind Maven took notice that there is a problem and working on something.

The key is, you don't start fighting with Maven.

"The philosophy of Tai Chi Chuan is that if one uses hardness to resist violent force, then both sides are certain to be injured at least to some degree. Such injury, according to tai chi theory, is a natural consequence of meeting brute force with brute force. Instead, students are taught not to directly fight or resist an incoming force, but to meet it in softness and follow its motion while remaining in physical contact until the incoming force of attack exhausts itself or can be safely redirected, meeting yang with yin." (http:// en.wikipedia.org/wiki/Tai_chi_chuan)

If you fight with Maven, you will lose. If you meet it in softness and follow its motion you will attain build enlightenment...

After about a year of experience with Maven, I find that:

* The "surefire" plugin is named ironically. The build often fails without a message about what unit test failed.

* Maven is extremely slow. Eclipse builds the source in 30 seconds max; Maven takes 5 minutes.

* Maven's output is extremely verbose, but if anything goes wrong, it "helpfully" tells you to rerun it with -X -e to get the actual error message and stack trace. You can run it with -X -e all the time if you want, but I think it generates so much output that it actually slows down the build. Yes, really.

* There are a lot of things you just CANNOT DO in Maven. One example is bundling a Maven plugin together with the rest of your Maven source and use that same plugin elsewhere in the project. It fails during the early phases of the default lifecycle because the plugin isn't installed yet. But only by getting to the later phases of the lifecycle can you install the plugin. But you can't do that because the early phases fail. An infinite cycle of despair is right.

* Even the most trivial project takes a few screenfuls of XML to set up in Maven.

* Maven's dependency management is poor, too. If you have some library A that pulls in B, and another library C that pulls in a slightly different version of B, you can get misbuilds. It's your job to identify and fix these problems. They will show up at runtime. And let's not even talk about SNAPSHOT and how horribly that can go wrong.

* Maven is also known for pulling in tons of unnecessary dependencies. But hey, you don't mind 50 more megabytes of jars on your CLASSPATH, do you?

I was honestly expecting the article to address some of these points, instead of vague philosophical ramblings. Seriously, Maven is horrible. This article gives you no inkling of just how bad it is.

That said, I recently had to use ant after many years doing Ruby, and I was surprised to find how many old annoyances were still there, like scarce/painful support for conditional execution. I wanted a task to `cp` in development and `scp/rsync` in production, and I wound up having to duplicate a lot of code. Also I'd have thought that someone would have added the copy task's filter support to scp by now. (If anyone knows a clean way to do any of this, I'd love to hear it.)

However we do not yet seem to plug http in the same way we pipe command lines. Something seems to be missing,

We use Maven with Bamboo for continuous integration and it just works. We play with the idea to move to Gradle b/c it's more powerful, but the effort is non trivial and the real benefits - beside the 'but it's programmable' - not clear.

Your plugin has to play by the rules, which is a good thing.

Maven's useful for more than the beginning of a project. If you're working on real-world software, you're probably not the only person who has to understand your bespoke build, if that's the way that you've chosen to go.

  A built tool should allow (when needed) the full 
  flexibility of a Turing complete programming language. 
  Because eventually your project is probably going to need 
  it.

It uses the groovy language so hopefully it doesn't fall into the same trap as described in the post.

If it has it, someone is going to think they need it. A year later you have 10,000 lines of Ant and nobody to maintain it.

Building code is not hard. Discipline is hard.

My only fear with maven is when things break or someone how you have a wrong version of maven that the script requires or downloads fail or something else. If the script fails you can't do anything. And who knows, the only that was required was just compiling the java source with minimal dependencies.

Here's a very short list of things that are massively obnoxious to do with Maven, but are perfectly reasonable:

- submitting code to a code review site like Gerrit.

- generating code (for example, a parsed SNMP MIB that you want as a Java class so you can refer to it easily).

- integration with a tool like Sonar (yes, there's a Maven plugin. You ever used it?).

- code coverage analysis (yes, there's a Maven plugin. You ever used it?).

- FindBugs style analysis (yes, there's a Maven plugin. You ever used it?).

- interesting dependencies on external libraries (to pick an example from Ruby, the json gem was horribly broken at 1.4.2 and generations of projects have varying requirements for json < 1.4.2, json > 1.4.2, and many other worse things.

- C code (through JNI or anything else, a perfectly reasonable thing to want to do).

- deployment.

From bitter personal experience it is possible to get so wrapped up in this that you think you have achieved something amazing when you finally finish, when it could have been done in a few hours in make or rake.

"Oh, but you can integrate it with your IDE!" The only thing your IDE actually needs to get from the Maven POM is to understand where your source code is, what you depend on and where it is, and how to run your tests. Everything else they do with command line calls, just like it was Ant.

e.g. take this selenium install instructions (http://seleniumhq.org/docs/03_webdriver.jsp)

Ruby:

    gem install selenium-webdriver

   <?xml version="1.0" encoding="UTF-8"?>
    <project xmlns="http://maven.apache.org/POM/4.0.0
                 xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance
                 xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd>
        <modelVersion>4.0.0</modelVersion>
        <groupId>MySel20Proj</groupId>
        <artifactId>MySel20Proj</artifactId>
        <version>1.0</version>
        <dependencies>
            <dependency>
                <groupId>org.seleniumhq.selenium</groupId>
                <artifactId>selenium-java</artifactId>
                <version>2.28.0</version>
            </dependency>
            <dependency>
                <groupId>com.opera</groupId>
                <artifactId>operadriver</artifactId>
            </dependency>
        </dependencies>
        <dependencyManagement>
            <dependencies>
                <dependency>
                    <groupId>com.opera</groupId>
                    <artifactId>operadriver</artifactId>
                    <version>1.1</version>
                    <exclusions>
                        <exclusion>
                               <groupId>org.seleniumhq.selenium</groupId>
                            <artifactId>selenium-remote-driver</artifactId>
                        </exclusion>
                    </exclusions>
                </dependency>
            </dependencies>
        </dependencyManagement>
   </project>

   mvn clean install

Let me illustrate what we do at work to explain. We use Jenkins as a build server and a hosted git repository. Every accepted pull request to mainline-dev is picked up by Jenkins, Jenkins runs mvn test on the commit, if the build passes, Jenkins integrates the change into mainline-stable, and Jenkins runs mvn deploy to get the latest build on a mainline-stable server.

Setting up this workflow required little customization because mvn's output was well-defined and designed to be composed within other environments, and jenkins is one common place that understands it. So are the IDEs at your work. So is the next person to tweak the environment who can look at docs to figure out what's going on. It's no different than how sed knows how to interpret tr's text output.

Does this workflow work for everyone and every language? No, not at all. Does mvn the tool work for everyone and every language? No. But I see it as an example how mvn, a build tool, fits into larger process and is a composable tool the author lauds.

P.S. The biggest benefit for our shop is twofold:

1) Dependencies get resolved automatically in all environments. Unlike my last gig, developers run the same version of libraries as the prod machines do.

2) Developers run the same test cases locally as jenkins does. We've only had one build break.

    And then the person that wrote them leaves and we're fucked.