Archive for the ‘Work in progress’ Category

A hot topic for the scalability of ”’Meandre”’ flows is the ability to provide distributed flow execution capabilities.  I’ve written some notes about how ”’Meandre”’ approaches this issue here.

Yup, we are running toward a release. No, I have not heard about the date yet, but we are getting ”’Meandre”’ ready for it.  Among other things, we are currently working on:

* Fixing known bugs.
* Preparing the tools to be easy to install and run.
* Finishing documenting the code and the user manual.
* Sketching a quick start guide and a tutorial.
* Preparing the website.

I guess we can say that something is coming 🙂

For the last week I have been working on 1.3 Xuixo (another wonderful pastry originated in my hometown, a naming convention I sometime use). The web service api is still the same, so you intrepid souls should still be able to use it without noticing the change (unless I added some funny bug–not unlikely).

The main changes are under the hood.  I refactored most of the code to remove the static dependencies.  Now you can start many self-contained MeandreServers in your embedded usage of Meandre.  This will simplify writing test cases against the web services api.  I have not moved the old (1.2) web services test, because it needs revamping.

I have also been working on adding new tools:

# Mau files are now self-contained. A mau file is a jar containing the repository rdf and all the contexts required to run (jars). This is usually generated via the ZigZag compiler or interpreter.
# Improved the ZigZag compiler to generate the new mau file format.
# Modified the ZigZag run time to be able to run the new mau files.
# Yes, you heard it right. I added a ZigZag interpreter.  It comes with a console and you can build flows incrementally, merge flows together, save flows and maus, run a flow, etc, among other things. Just type “help” and it will talk to you.

I finally gave up and learned ant.  Great news, I guess, for the rest of the people on the team:).  This version (1.3) includes my take on the old (1.2) build.xml.  It provides the following tasks:

* compile
* compile-tests
* clean
* run-tests
* test-reports
* javadoc
* dist

No, it does not provide a run task, or like capability. You should not run the Meandre server there. The dist task generates the usual libraries (and a few more I have added) and 4 standalone jars:

* meandre-server-X.Y.jar (Meandre server)
* zz-X.Y.jar (The ZigZag interpreter conole)
* zzc-X.Y.jar (The ZigZag compiler)
* zzre-X.Y.jar (The ZigZag runtime for MAU files)

You can run each of them by typing java -jar name.jar. So, for instance if you want to run the core, copy meandre-server-X.Y.jar to wherever you want to have the working directory and run java -jar meandre-server-X.Y.jar. That’s it: self contained and auto configured. You can stop it then and play with the config files if you want to tweak things–or instance use MySQL/Oracle instead of the embedded Derby.

As I said before, this is for intrepid souls.  Please let me know if you run into problems, bugs, etc. Also, jira now contains a Meandre-Terracota-Executor project; feel free to report the bugs there directly and I will do my best to catch up.

We’ve set up the main website for Meandre, the semantic-driven data-intensive flow engine. We are currently working to get you a few things: a first alpha release of the ”’Meandre”’ infrastructure, the means to build and run data-intensive flows, and enough documentation to get you jump-started 🙂  I will keep posting our basic activities toward these goals.