NOBLE APE INFORMATION - JUNE 2003
SEVEN YEARS AND APPLE
NOBLE APE SIMULATION 0.661 and 0.662
CORE USER BASE
INTRODUCTION
I have started this mailout early to try to include as much current information as possible. This month has seen two versions of the Simulation released. Apple engineers have been using the Noble Ape Simulation to high-light speed improvements with Mac OS X hardware and these engineers will display the Noble Ape Simulation at Session 506 of Apple's World Wide Developers' Conference at the end of this month.
This month's tight release schedule has also highlighted a core of Mac OS X users of the Simulation - roughly 1,500 - who update the Simulation with each version release, write about their use of the Simulation on websites and assist others in understanding the Simulation. I have encouraged those that correspond with me to join this mailout - although subscription to the mailout remains relatively steady.
SEVEN YEARS AND APPLE
Noble Ape celebrated seven years of development on 13 June 2003. This milestone has been marked with the display of the Noble Ape Simulation at Apple's WWDC by two Apple engineers, Nathan Slingerland and Sanjay Patel. In a hybridisation of open source and Apple's secrecy surrounding the event, I am not totally sure what the Noble Ape Simulation will be used for. I have seen one slide from the talk so-far. But the source code for the Simulation used at the event will be hosted on the Noble Ape website after the event.
In addition to WWDC, Apple will distribute a regularly updated version of the Noble Ape Simulation source code with their developer tools into the future. This brings a number of additional users and developers to the Simulation.
The features I have seen in the early source code provided by Apple, and my discussion with Nathan and Sanjay include threading (which doubles the Simulation speed on dual processor machines) and vectorisation (which improves the speed on G4 machines). Both these features are applicable with some conversion to other processors. Like the development of the Generic Platform Interface for user interface features, a similar thin interface for threading and vectorisation could be used to adapt these features for Windows and Linux.
The challenge now the Simulation has been adapted and distributed by Apple is to ensure the other platforms of the Simulation are maintained at a similar level. Maintaining an OS independent Simulation Core and GUI layer is central to the requirements of the Simulation. The source code I have seen from Apple breaks this currently - although I am confident with a bit of time and development these paradigms can be re-introduced and the Apple source can be folded into the active development.
NOBLE APE SIMULATION 0.661 and 0.662
Two Mac Carbon specific versions of the Noble Ape Simulation were released this month. The first (0.661) attempted to introduce full file handling and eliminate the Proof-of-Concept status of the Carbon version. For the past three releases, the Carbon version has been the centrally directed version for Mac downloads. It is important to have a full version so all the Mac downloads can experience the full Simulation. The idea of Proof-of-Concept has been remove and replaced with the GPI in all future versions. Implementing full file/OS functionality in one step from the GPI will be the new development ideology.
0.661 was released in a tight time frame allowing only two weeks from 0.660. Unfortunately it wasn't tested enough on different System versions and 0.662 was released a week later. 0.662 fixed the final file bug and introduced integer only processing in the GUI layer of the Simulation. The Simulation Core still uses floating point although the higher resolution development in 0.659 and 0.660 eliminates the need for floating point in the future.
With the tight time frame on the past three releases, I am planning on releasing 0.663 in the middle of next month. I intend it to be a full release on all platforms including Mac Classic and Windows. Following the release of 0.662, I heard from Mridul P that he is continuing the Linux development towards a GPI version.
CORE USER BASE
As mentioned in the introduction, a core group of Simulation users is emerging. What interests me about this user base is they refuse to be tracked through normal means. When I started the mailout - almost at the same time I started the Nervana Project in 1996 - I took a survey of those on the Mailout to investigate what they wanted to see in the Simulation and the Project's development over the following year or so. I can't remember the specifics of the Survey although I remember receiving a roughly 80% response rate.
I would like the development of the Noble Ape Simulation to be responsive to the needs of the user base. Although each release receives roughly 1000 downloads in the first day, through the entire release I rarely receive emails from end users.
So how do we track the regular downloaders?
I have thought that a neutral survey form with a standard free survey company should do the trick. If I can get information in percentage terms - per user - and develop the Simulation based on that information, so much the better. I have two thoughts - either embed the survey in the downloaded release or embed the survey on the site. Either way - one user, one survey.
A request I don't need to survey on - more information. Users want more information. They want greater detail and things explained simply in other sections. Describing the cognitive and biological simulation to Apple, I heard myself giving concise paragraph answers to the questions of the cognitive and biological simulation that should be offered in the basic manual distributed with the Simulation.
Hope all is well with you all,
Tom Barbalet, 15 June 2003.