SourceCloud of the Debug Visualisation Project

However, it is interesting, that the strings ‘end_of_line‘ and ‘sp_cleanup‘ found their way in. Luckily, only a few one-character names and numbers are present, meaning, we managed to created longer variable names…

The cloud of the VIATRA2 model transformation framework is more interesting: it is a much larger codebase, developed for several year by around a dozen people. Because of the size I generated the cloud using 300 words, but almost no Java keywords found their way into the graph.

SourceCloud of the VIATRA2 framework

However, the mandatory comment parts did, e.g. nbsp or the names of some creators. A lot of EMF-specific keywords are also present. They are present because we have quite a large EMF metamodel that provides several hundred generated Java files.

At last, but not least I also checked the cloud of the EMF-IncQuery project. This is smaller, then the VIATRA2 project, but larger then the Debug Visualisation. Additionally, it has a shorter history, but with several key committers.

SourceCloud of the EMF-IncQuery project

My personal favorite is this diagram, as it shows a lot of nice things: e.g. four digit numbers in large quantities (magic numbers FTW) or the absolute champion string: ‘HUKfyM7ahfg‘. All of them are present in the Ecore Diagram of the defined metamodel (the meaningless string 152 times as a postfix of an identifier ). At least, I learned something about the internal model representation of GMF.

And how do your projects look like?

This release is a bugfix release only, it is avaiable from the update site or Eclipse Marketplace, and recommended for every user of the plug-in.

Two minor new features were also introduced: it is possible to turn off name trimming, and the values of variables is now displayed as a hover.

I created a screencast to show these new features (and also some other, as the last screencast was created over a year ago.

(If the embedded video does not show in your feed reader, you can watch it on YouTube.)

The changes are three-folds in this release: first, a view model is defined for the visualisation. This one is only interesting for us, developers, as it enables us to store calculated data together with the referenced variable or value, making it easier to produce new features in later releases.

The other changes are more visible: we integrated our approach with the debug platform: it is possible to add variables to the visualisation by using the context menu of the Variables View. For similar reasons, our custom filter implementation was replaced with the platform-defined logical structures. This means, more specific filters are available, and it is possible (at least for Java) for the user to define more in the corresponding Eclipse preferences page (Java/Debug/Logical Structures).

These stuctures provide a way to removed some (possibly) irrelevant variables from the visualisation, as it is possible to define on a per-type basis the relevant structure (and it is also possible to define multiple structures for the same type, allowing the user to choose between them).

These structures can be selected from the context menu of the values in the Visualisation view. Such a possible selection is depicted on the following screenshot:

Logical structure selection for an ArrayList

Warning: the logical structure selection is (and possibly will) not synchronized with the variables view.

Finally, some internal changes were made to ensure, the plug-in is easier to use, such as the ability to define custom orderings for some layouts.

For a list of all fixed issues visit our issue tracker.

The new release can be downloaded from our update site right now (http://eclipse.cubussapiens.hu), and hopefully shortly also from the Eclipse Marketplace.

Update: Detailed project information is available from our Google Code project page.

Update2: The new release is available (and installable) from the Eclipse Marketplace.

I’m proud to announce that after a long time, we finally managed to throw out a new release. I’ve managed to improve the core functionality therefore this release contains a significant advance in overall performance.

Also, Stampie managed to fix a long-awaited feature, the two-way synchronization with the variables view.

In the future, we plan to redesign the use-cases of the tool. Currently, we’re in an experimental phase, which means we do not know how we want it to work.

This includes some minor adjustments in order to provide a more seamless user experience (sorry for the bullshit, but I just couldn’t withstand it ).

Well, the new release fixes a nullpointer exception, and adds an option to save the current graph.

There are further tweaking regarding usability: the default layout is changed to Tree Layout in order to allow faster responses, and it is possible to hide the Local context node. These changes are the beginning of some serious overhaul, our goal is to increase the usability of the plugin.

The new release is called 0.7.0, as it introduces major new features:

• Preliminary support for Filtering nodes: similar to the Logical Structures feature it is possible to define the structure, and only visualise that (e.g. in case of Lists only the content should be depicted, not the other attributes).
• Drill in: it is possible to define a new top-level element instead of the Local Context, and hide other items.

There are also some minor updates, such as the reintroduction of the Simulated cooling layout algorithm.

On the other hand most changes are not visible in the user interface, but the revamped code should allow us to create some new features in the next service releases.

We recommend the update for everybody. Some technical information:

• After some fighting with the update site project the category information is lost. I have no idea of the cause.
• There is a new feature present in the update site: the Java Filter project defines filters for some items in the collections API.

I’m satisfied with the currently used approach (for now), as I’ve managed to separate different functionalities, and connect them in a simple way. Also, it is reasonably faster than before, but it still makes the eclipse lagging when the user steps the debug context, which I don’t know why (yet). Anyway, navigating in the graph, hiding/showing nodes and other display-related functions are now fast enough.

For a more detailed explanation of the new operation, see the wiki page InputPackage.

ps.: for a change as big as this, I’m thinking of a maintenance step in the version numbering (0.7.0 instead of 0.6.2), what do you think? I’m hesitating because of the lack of new features.

Version 0.6.0 introduced some serious regressions from the previous versions, so we decided to create a quick service release.

This release contains fixes for Stack Overflow and Hidden nodes issues, and added some minor new features as the ability to zoom the graph.

You can use P2 to automatically update your installation, or install the new version from the http://eclipse.cubussapiens.hu update site.

We are hoping, this release will be usable than the previous one.

It was a while ago I created this introductory screencast (it used a modified version of 0.5.1 (perhaps it can be considered as a never-released first alpha of 0.5.2), so it is not totally correct but it helps a lot to understand how this can be used.

If not, please don’t hesitate to inform us how to correct it.