Code Recommenders: Developers come and go but the code remains

Session Type: 
Standard [35 minutes]

Do you know how long it takes for a new developer to become productive? And do you really know what the developer you let go just last week was working?

While developers come and go, their code remains. And with it a wealth of implicit knowledge that is just waiting to be tapped. Using the technology behind Eclipse Code Recommenders, we will demonstrate what information can be extracted from code in open‐source software repositories to support ongoing knowledge transfer between developers.

By the end of this session, you’ll have a clear understanding of how this technology can be applied to other repositories and of the workflow required to transform your projects’ private codebase into your projects’ knowledge base. Give your development teams a head start!

Schedule info

Time slot: 
26 March 16:15 - 16:50


Experience level: 

Re: Does Code Recommenders work offline?

This question was submitted using the (anonymous) feedback form: "What if you don't have access to the internet? Could this work on an internal network?"

I'll attempt an answer here, as I obviously cannot do so by e-mail: Absolutely. Code Recommenders also works with locally stored models. In fact, during the demo part of the talk, the JavaFX models I mined where stored in my home directory. All I needed to do was enter a file: URI in Preferences/Code Recommenders/Models. Of course, you can also share your models with your colleagues by putting them onto an HTTP server. But even in the latter case, Code Recommenders will maintain a cache of downloaded models below you workspace's .metadata directory, so your code completion will stay intelligent even if you are offline for a while.

That being said, why don't you follow the tutorial that's part of the Devleoper Preview? It explain the whole workflow in just five minutes.

Copyright © 2013 The Eclipse Foundation. All Rights Reserved.