notes.txt

Folder Structure
-----
Processing Plugin for Eclipse: the "feature" plugin, an xml document
describing version numbers, component plugins, copyright and licensing
info, etc, that wraps everything into one nice bundle for users to
install.

processing.plugin.appBundle: made with the "plugin from existing jars"
wizard. contains the processing.app code that I could directly reuse,
and its dependent code. the version number indicates what version of
the p5 editor the jars came from

processing.plugin.core: non-ui functionality. has a builder, the stub
of the exporter, and a simple model for libraries. Also contains a
static utility class of methods copied from the old Base and PApplet
classes that were needed, but couldn't be reused without invoking
swing UI that made Eclipse grumpy.

processing.plugin.ui: all of the ui functionality. The editor, syntax
highlighting, related utilities, launching (runner), the wizards, and
the perspective are here.

processing.plugin.updateSite contains some xml used by the Eclipse
update and installer tools. It also contains compiled versions of the
plugins for distribution, and opening site.xml in Eclipse is where the
build happens.

Project Lifecycle
-----
On creation or import, Processing sketch projects have a 'nature'
attached to their metadata. The nature attaches a builder and sets up
a Java project. When a change takes place in Eclipse through the UI, a
delta object is passed to all the projects and builders. Thing of this
delta like a diff output. In a Java project, the JDT figures out only
what needs to be recompiled based on each change. Processing lacks the
functionality to recompile individual segments of sketches, so instead
we interrogate the delta to make sure it is resulting in a change that
will affect our project before initiating a build. The overwhelming
majority of deltas are ignored because they don't affect compilation,
but we still end up frequently rebuilding the whole sketch.

When the builder runs it checks for preprocessor errors, and if any
are found it marks up the document accordingly and bails. If none are
found, it puts all the preprocessed files into a JDT project managed
internally, and lets the JDT take over compilation.