Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[cdi-dev] CDI-lite TCK meeting notes

CDI-lite TCK Meeting Notes


Scott M Stark

Dmitry Kornilov

Graeme Keith Rocher

Gurunandan Rao

Matej Novotny

Tomas Langer


We had a quick meeting today to discuss spinning up a CDI-lite TCK.


Expectations:

  1. The existing Arquillian/TestNG unit testing framework will be used.

  2. An additional CDI-lite source processor SPI will be added to callout to the implementation being tested to perform build time processing for each unit test deployment.

  3. The test sources are already available as a source artifact and as part of the TCK distribution bundle. The CDI-lite TCK instructions will have these sources extracted locally and this location will be used to identify the source files associated with each unit test deployment.


Questions:


  1. Base archive format is a jar, assume war is no longer suitable?

    1. Yes, jar is expected base format

  2. Leverage types from the javax.tools.JavaCompiler for the source processor SPI?

    1. Yes, the expectation is that there is a TCK SPI that provides the classes and associated java source files to the implementation being tested for each unit test deployment. This will return classes (or overwrite them on the compilation path?) as well as any build time errors using the javax.tools.Diagnostic interface. These will be mapped to CDI deployment exceptions that the unit test framework will see for matching with expected exceptions.

  3. Create new TCK based off of the current embeddable tests?

    1. Yes, we expect to start from the current https://github.com/eclipse-ee4j/cdi-tck master branch where the tests have been separated into a base set of embeddable tests that have no downstream Jakarta API dependencies and this will be the base set of tests feeding into CDI-lite. We may need additional TestNG test groups to filter out some of these tests, but at this point what CDI-lite inappropriate tests exist are unknown.

    2. Scott expects that an initial prototype in some form should be available in a branch off of the master cdi-tck repo by the June 15 meeting.


Back to the top