Maven build

[chrido]

For build and unit testing purposes, I migrated the build of the java project from ant to maven.

Would you accept a PR?

[gtoubassi]

Sure I'd like to take it. Send my way. I'd also be curious to know what you are doing with femtozip (compressing small payloads or sdch dictionary generation).

[chrido]

We are currently experimenting with reducing bandwidth in pub/sub, many small similar payloads.

Here is the PR: #10

[gtoubassi]

Hi pardon my ignorance with maven. I patched in the pull request, and ran install-local.sh. It succeeded but didn't appear to install anything. I get a file in my local maven repository that is a jar with no class files:

Archive: /Users/gtoubassi/.m2/repository/org/toubassi/femtozip/1.0/femtozip-1.0.jar
Length Date Time Name

---

    0  12-07-15 19:14   META-INF/
  174  12-07-15 19:14   META-INF/MANIFEST.MF
    0  12-07-15 19:14   META-INF/maven/
    0  12-07-15 19:14   META-INF/maven/java-femtozip/
    0  12-07-15 19:14   META-INF/maven/java-femtozip/femtozip/
 3431  12-07-15 18:59   META-INF/maven/java-femtozip/femtozip/pom.xml
  104  12-07-15 19:14   META-INF/maven/java-femtozip/femtozip/pom.properties

---

 3709                   7 files

I'm sure this is ignorance on my part. Is install-local.sh supposed to actually build the class files? If I do mvn package I get class files built. The jni tests fail which is understandable given I don't have the shared library built.

So if I do:

mvn -Dmaven.test.failure.ignore=true install

I get the jar with classes installed.

So I guess my net question is whats the point of the install-local.sh if it doesn't install a useful jar in the local repo?

[chrido]

Yes, that's not very useful. Currently it creates a jar and then installs it using a different command. I'll correct the PR tomorrow.

[chrido]

Sorry for the delay, busy times. I changed the groupID to org.toubassi.femtozip, now mvn install should install it with the correct name in the local maven, no script necessary any more.

[gtoubassi]

I merged the PR. Thank you.

Can you tell me more about your use case? Do you have any sample data you're willing to share that you are using to benchmark? I have some ideas for improving dictionary generation and would be interested to see if they improve your case.

[chrido]

We don't have a "real" use case, more in research; we want to demonstrate how it can be used in Content based Pub/Sub, where each message is matched with a number of rules. We have some test data sets where we looked at the sampling interval and length of the interval within a continuous message stream since this is our use case. This could be extended for other use cases too, to have multiple bench-marking data sets. We can share this. My email is doblande [at] in.tum.de