Creating an Eclipse product from KNIME

Hi all,

I am trying to create an Eclipse product configuration from the KNIME SDK. I intend to distribute my node extensions as a product so that users do not have to learn update sites and could just download it. In one of the extension nodes I work on, R bundles for 32 or 64 bit Windows are provided in the update site to facilitate the installation of R. Many, many experienced users fail to configure R in KNIME. I would like to create one product for each architecture so that the users could download a vanilla KNIME with the extensions and the appropiate R bundle and would not need to configure anything.

In the minimal example from the SDK setup I am just creating a product configuration file into an empty project (product) and initializing the product from the KNIME launch configuration.

This creates a product file using KNIME Product and KNIME Application.

Unfortunately, I am getting errors related to missing KNIME Product. Even when I try to create a dummy product with the button New (second picture) many errors appear due to missing KNIME features which are added in the Contents tab of the product.

I am sure I am making more than a couple of errors here and would appreciate any help.

Lots of thanks,

Hi Miguel,

I can’t help you with your Eclipse problem, but there’s also a different approach to solve this. If I understood you correctly, you want to build pre-configured KNIME client that you can roll out to your users. This includes some custom extensions, and also needs to be shipped with R binaries.

There’s two points to consider here. First, you can create the actual client by just configuring a reference installation with all needed extensions. You can then zip the whole Analytics Platform installation directory and distribute it to your users. All they need to do is unzip.
However, since the preferences for R etc. are kept in the workspace rather than the installation folder, you need a second piece here. If you have KNIME Server, you can use it for centralized management of client preferences. This means that you can store a number of different preference profiles on the server, so that each client just pulls the correct preferences upon startup. That way, users won’t have to set anything manually.

Is this something that would help in your case?


Roland is right, I also cannot recommend building a product on your own. Instead use the existing KNIME Analytics Platform and install your extension(s) into it.

Hi Roland and Thorsten,

thanks for your answers. I have created in the past bundles like you said by zipping a KNIME installation. However, since I have a CI pipeline that also deploys releases I wanted to create the bundles there.

Thanks a lot,