|
Hi Jason, In addition to Willie's update to your history, there is a second branch to this family tree. At roughly the same time as the Sun/DEC/GATech work on Mercator, the Disability Action Committee for X, and the implementation of RAP, there was a group of Macintosh AT vendors gathering together to try to improve the state of Macintosh accessibility. More or less led by Berkeley Systems (of inLARGE and outSPOKEN fame), we tried to build an API for Macintosh accessibility to convey information about components - text, controls, etc., to AT on the Mac so that it wouldn't need to patch QuickDraw and use all the various other hacks to get information for screen reading and other AT uses. Two particular problems we were trying to solve were the inaccessibility of HyperCard, and problems with X Windows servers on Macintosh. In addition to screen reading/magnification, we were looking at some interesting AT uses cases coming out of the Madenta ScreenDoors application - switch based control of the desktop & mouse where UI controls had "gravity" that would attract the mouse when moving nearby under switch control. We took these nascent ideas to Apple, and presented them under the banner of "AccessAware" at a World Wide Developers' Conference in 1993. Unfortunately at that time Apple didn't have the energy or vision to take this up, and without the backing of the OS vendor, the idea fizzled out. Meanwhile at Berkeley Systems, I started attending the early Disability Action Committee on X meetings - where I had the pleasure of meeting Willie Walker and Earl Johnson - and we hired a summer intern who produced a prototype port of our outSPOKEN for Windows codebase to X Windows. The prototype behaved well in things like xterm and the simple mail application & calculator, but was utterly useless with things like Framemaker because Framemaker rendered its own text (rather than using the X text rendering calls) - underscoring the need for an API (whether RAP or something else - like ATK/AT-SPI) that apps could use to communicate what they were doing while being free to use whatever techniques they wanted to paint to the screen. We shopped the prototype around to potential government customers, but didn't find enough of a potential market to justify the work to release a product. Likewise our early investigations into potential grant funding for an outSPOKEN for X failed to pan out. Meanwhile, though the Apple AccessAware work fizzled out, Microsoft started taking an interest in helping AT vendors, and I was an early consultant on what became MSAA (much to my disappointment). In 1995 Microsoft's AT efforts picked up in a big way coming out of Massachusetts' embargo of all Microsoft products due to the inaccessibility of Windows 95 (see http://blogs.sun.com/korn/entry/massachusetts_open_document_and_accessibility#Massachusetts_Accessibility_Background for a writeup of that event), which is where much of the MSAA energy came from. Microsoft also licensed the JAWS off-screen model, but then didn't do anything with it. They also started a project with ONCE and others to build a set of screen reader utilities - rather similar to the GUI Access Toolkit that Berkeley Systems had and used for outSPOKEN windows and was licensing to screen reader vendors - but that too never proved useful in the market. By 1996 most of the Berkeley Systems folks were pretty tired of fighting in the screen reader trenches - spending 80% of your time reverse-engineering successive releases of Windows and perhaps finding a bit of time to squeeze in a couple of new features around the edges - and we sold the technology to ALVA BV in The Netherlands. And that's when Earl Johnson hired me (along with Willie) to work on the Java Accessibility API, which as Willie notes led to AT-SPI among other things (like the UNO Accessibility API of OpenOffice.org, IAccessible2, and even WAI ARIA). Regards, Peter Korn Accessibility Architect, Sun Microsystems, Inc. Hey Jason: You are mostly accurate here. :-) The only thing I'd change is that UltraSonix was originally the Mercator project, which was a joint collaboration between Sun, Georgia Tech, and DEC. The public work really began around 1992 when we formed DACX, the Disability Action Committee for X. The influence of RAP (the service-based API that Mercator used) on today's designs is for real. When I was hired by Earl Johnson to work on the Java Accessibility API in January 1997, I applied the concepts of RAP to the work. :-) The Java Accessibility API ideas then fed into other areas, such as AT-SPI. If I have one MAJOR regret for this whole thing, it's that we didn't provide a good specification for the event model in RAP. This not only includes specifying the event types, but also the ordering of them. In addition, I wish we had devised a way to specify a grouping of events based upon a single user action (e.g., a keypress). With a tighter specification, I think a fair amount of complexity in the screen reader might go away. Will Jason White wrote: |