Open Topics

Here you can find:

• information about topics which were started, but not completed
• ideas and implementations which were never integrated into the main branch, but might be interesting.

Therefore this page can be used for finding open and interesting topics which can be continued. Please, let us know (e.g. via our Gitter or mailing list) before you continue a topic so we can share our thoughts and experiences.

Instant Session Start

Status: Currently the author (@stefaus) of the feature wants to finish the rest of this work and some patches are already available. Nevertheless the topic is on hold mainly because it is mandatory to change the way resources are shared first (from project based to per resource tracking).

Goal of this topic is to speed up the initial file sharing process at session starts. Instead of waiting for full synchronization, this feature prioritizes files and allows direct access after receiving. This main goal has already been achieved, but users are still bound to the read-only mode during project sharing and optimizations are open / on hold for merging. More information can be found in this thesis (German only).

An Overview of the current work state is documented here: Project Board: Instant Session Start Feature and a broader view at Project Board: Session Start Topics.

GIT Support

Status: The topic is open and it is likely that we will improve the git integration in the future.

Corresponding Pull Requests:

• Add JGit facade
• Add Activities to Share Commit

A former contributor started an implementation of a basic git support. The main idea was to send differences between the git history of two developers as git bundle. The determination of the differences was implemented via Saros activities. See page 18 of the the corresponding thesis for a corresponding interaction diagram.

Improving the Runtime of the STF Tests

Status: The topic is open, the stabilization of the STF tests is currently in focus.

A former contributor already thought about ways to reduce the runtime of the STF tests (which is is very high). She proposed the following approaches (see the corresponding thesis for more information):

Conditional Testing

See here for the corresponding Pull Request
The main idea is to skip further tests if a basic test already failed.

We did not integrate the pull request, because:

• The approach requires to model a dependency relation between tests.
  • The dependency relation model was not compile-save.
  • The dependency relation model was redundant, because it was implemented via a direct relation and the test order within the test suite.
• We don’t know how much time is saved by this approach in practice.
• We want to avoid additional complexity which hampers the induction of new contributors.

Minimize the Setup/Teardown Effort of Similar Tests

See here for the corresponding Pull Request
The main idea is to identify tests with similar setups. Instead of recreating the setup for each test, the old setup is just cleaned up.

We did not integrate the pull request, because of:

• The approach introduces a lot of complexity and adding a new test becomes very complex.
• The implementation introduces different tear-down methods which has to be changed manually in order to execute tests separately (not in the context of the test suite).

Deprecated Approaches

Whiteboard

Status: Removed from the master branch. You can find all information about the idea of the Whiteboard here.

HTML-GUI

Status: Removed from the master branch. You can find all information about the idea of the HTML-GUI here.

Java-JavaScript Bridge

Status: Irrelevant, because of the deprecation of the HTML-GUI.

Corresponding Pull Requests: Reimplementation of the ui classes in Kotlin:

• Part 1
• Part 2
• Part 3
• Part 4
• Part 5
• Part 6
• Part 7
• Part 8
• Part 9
• Part 10

The main idea is to reduce redundancy and complexity during the interaction of Java and JavaScript (in the deprecated HTML-GUI). Therefore this work compared different approaches (Scala with Scala.js, Kotlin and Fantom) which allow to generate the JavaScript code based on a JVM language. See this thesis (German only) for more information.

We did not integrate the pull requests, because:

• The approach introduces a new language and therefore increases the complexity of our project and build process.
• The savings of redundancy are smaller than expected. Only the redundancy of the model classes can the avoided.
• The most interesting part (The injection of java functions into JavaScript and vice versa) is not simplified by the approach.

Enhance the STF for the HTML GUI

Status: The specific approach is irrelevant, because of the deprecation of the HTML-GUI, but the main idea (IDE independent STF Tests) is still important.

See here for the corresponding Pull Request
As already implemented for SWT components a new view is introduced which allows to trigger GUI events in the RMI server.

The pull request contains a basic implementation with mostly stubbed methods.