Thursday, July 24, 2008
Finally some documentation on the "rest of Field" — the new palette window.
Tuesday, July 22, 2008
The number of people using Field has grown to the point where Marc's having a hard time keeping track of everyone. A new Google group seems in order: http://groups.google.com:80/group/field-development/ please join so we can consolidate our efforts.
Also, the documentation sprint begins: first, SwingIntegration — how to integrate a little Swing into the Field environment; second — updated ScalaSupport notes.
Friday, July 16, 2008
In preparation for three Field related projects this summer, Marc's just committed the results of three weeks of on-the-road laptop Field hacking. These changes, from large to small: Field now uses the "alpha 0" of Jython 2.5 — although a shade scary, the Jython people could probably do with the bug hunting themselves. In exchange we get two features that Field has been waiting for for literally years: generator expressions and python decorators. New hooks into the auto-completion in the text editor allow for integration of documentation from various sources (of course, now we need some documentation). A new component class allows Swing components to be drawn and interacted with directly on the canvas — including text editors and complex swing components. Swing with the right Python wrappers can actually be quite pleasant and terse. There have been several small fixes to the Scala Plugin which is now useful enough for Marc to at least begin to learn Scala properly. Finally, Field now registers itself as a .py file editor and automatically imports or refreshes .py files into workspaces. I think we are doing a good job of moving towards a widely useful and documented-enough binary release by the end of the summer.
Sunday, June 15, 2008
Goaded by the meetings in London last week, we've completely overhauled the window and "workspace" management in Field. It's always been poorly tested by us — with our use cases seeming to never overlap with everybody elses — and a source of confusion. In any case, the workspace manager is gone; an honest File menu has arrived and tool palettes are compressed into a single window. Everything already feels much more solid, it will be hitting the repositories tomorrow.
Thursday, June 12, 2008
Just pushed a medium size commit to the main repository that covers a number of small niggling bugs that were fleshed out in a "Field meeting" in London last week. OpenEnded has been collaborating with Random Dance / Wayne MacGregor as part of the "Entity" project, and we've been kicking Field into shape as a platform for a very personal idiosyncratic "choreographer's sketchbook". Expect a) some interesting text parsing features to be added to Field, b) some actual documentation of Field's underlying graphics system and c) a more complete set of Scala features to happen this summer
Thursday, May 28, 2008
As work comes to a conclusion on Housebound, we've taken the opportunity to revise the Loops public release. The latest (very large) commit to the Field repository reflects hopefully the final refactoring of Loops into Loops specific code (available as a snapshot here) and code that's more general purpose. But above all, it contains the tweaks to the Loops renderer (which is essentially a different combination of the techniques behind Recovered Light and Point A->B) to get it to run on the new NVidia 8800 graphics cards from Apple. Hunting for the bugs in Apple's OpenGL stack was painful enough that I'm pretty close to saying that all of our interesting rendering is only going to be supported on that card form now on.
Thursday, May 1, 2008
Hard at work on the workflow for Housebound — OpenEnded's upcoming short stereo-film. En route, we've found a rather nice, short, self contained, example of Field as a "tool to build tools"— in this tutorial we show how the drawing and text editors can be, with a little code, turned into an experimental color-matching suite. I'm also posting my short code for doing "healing brush" like automatic in-painting.
Tuesday, April 22, 2008
A short page describing Scala-based success. After much Classloading trouble, the early stage of integration now works.
Monday, April 16, 2008
Spent the day perturbed from my true, stated course of documenting recent work, by an intoxicating possibility that adding support for the Scala language to Field might be quite straightforward. While it lacks a JSR-233 compatible scripting interface (and it's pretty clear that to those driving Scala this is low priority), it's Java integration is at least theoretically tight and we've wrestled its evaluation loop into the kind of state that we need it in. Thus we have a working prototype of a block transform that allows you to write Scala code directly in Field's text editor. It's missing nearly everything (most of completion, syntax coloring, embedded GUI elements and a working _self magic variable) but it is still somehow compelling. Scala is a language that has a really different perspective than Python, and hopefully it will both initiate a cleanup and re-factoring of the text editor code and extend the audience for Field.
Monday, April 14, 2008
A lot's been happing in the last month that's poorly reflected in this news log. Firstly the history format for "versioned properties" seems to have stabilized and the features that depend on this are all working much better. In particular history for copied boxes and pieces of code is now being correctly stored and analyzed. This means that you can write a useful piece of code, copy it to a number of places, change it here and there and then gather all of these multiple edits together together to examine them. But the simplest and most obvious win from this is that you can easily switch back and forth in time for a whole sheet, or a subset of boxes, using the history window.
Also there's a host of cosmetic fixes to the line drawing and text systems, but also some real feature additions, including a lightweight constraint system for expressing constraints between nodes on lines that harks back, in a way that I find very appealing, to Sutherland's Sketchpad system (now that's a dissertation).
Under the hood, we've begun to transition to Apple's new Mac Pro's and their NVidia 8800's — much more impressive GPU's than the previous ATI 1900's we've been using — so in the underlying graphics system for Field there's support there for Geometry Shaders. People with this setup should be warned: there seem to be some significant Apple bugs associated with doing intensive graphics on these cards while monitors have been rotated in system-preferences. Currently Loops suffers from some serious visual artifacts on this card, but we have a workaround implemented, we just need to back port it.
Finally, ScrubberPlugin has been checked in. It's designed to let you add devices that move boxes around, most notably the Connexion SpaceNavigator's that we use at OpenEnded. It's pretty straightforward right now, but ultimately there will be some record and playback features added to it as well. This takes Field one step towards being a live performance system rather than just a live "rehearsal" system.
Our next task for the next week or so returns to this site: documentation and examples.
Sunday, March 17, 2008
While the "history" features of Field are coming along, we've been cleaning and documenting a whole host of "small things" and "small bugs". The embeddable-gui tricks are working better than ever (and not just in the text editor: EmbeddingGuiOnCanvas) and we seem to have solved the copy and pasting problems.
Friday, March 14, 2008
The simple Core Graphics bridge has been committed and documented here. The BasicDrawing documentation is also having a little clean-up, to highlight the PDF export options and articulate more clearly how properties work.
Next: now is the time to spend a few days cleaning up the "history" features of Field (some of which remain damaged from the switch from SVN to Mercurial).
Thursday, March 13, 2008
Marc has just posted a .tar.zip file of the source for Loops, as shown by OpenEnded Group on the Feb 26th. We'd still like to give it some polish, and some documentation, and some package refactoring before checking it into a repository parallel to field and workspace. But as it stands it contains too much code that might be of interest to other people for us to feel comfortable continuing to hoard it.
Wednesday, March 12, 2008
In preparation for OpenEnded's new stereo-film series Depth, Marc has been bridging Apple's Core Graphics framework into Field. This augments BasicDrawing with GPU accelerated image processing. This ends up looking a lot like the Core Graphics functionality of NodeBox — except we encourage CIKernel's to be written directly inside Field's text editor itself. Expect a preliminary check-in, and a tutorial, around the end of the week.
Monday, March 10, 2008
A good first draft of the ProcessingPlugin has been added to the main repository (inside the /extras directory). We've written a tutorial to help people move over from the Processing world to Field here: SimpleProcessingTutorial. While there are still a few rough edges, as far as we can tell it's ready for some serious Processing experts to come in and give it a good test.
Friday, February 29, 2008
OpenEnded is recovering from the press conference / premier of the Loops Preservation Project with the Merce Cunningham Dance Company. After we've cleaned up the source code to Loops, we'll be posting it (probably as a separate .tar.gz) available here . Already there's a considerable (and considerably interesting) chunk of motion capture and video documentation of Merce's 1999 performance.
Other changes
There are other ways of tracking Field's development. Consider the Index-By-Date page or the timeline link.