Operating system interfaces are what you bump into…

Operating system interfaces are what you bump into when you are trying to do your work. There is no need for workaday users to see an operating system interface at all; the current OS interface is bureaucratic bloat, an unnecessary impediment and tollgate. Users are interested in direct links to documents, not in operating systems and aps. Opening screens should show documents, not an OS. The metaphor for the interface should be the information, not an OS, not an ap, not a marketing experience. – Edward Tufte

Operating system interfaces are what you bump into…

Operating system interfaces are what you bump into when you are trying to do your work. There is no need for workaday users to see an operating system interface at all; the current OS interface is bureaucratic bloat, an unnecessary impediment and tollgate. Users are interested in direct links to documents, not in operating systems and aps. Opening screens should show documents, not an OS. The metaphor for the interface should be the information, not an OS, not an ap, not a marketing experience. – Edward Tufte

Moving on

FSMLabs is about to make a major change in direction – with a focus on “enterprise” computing. Some of what we plan to do is brand new – for example, we have new ways of speeding up “black-box” trading in financial markets. Some is old – Cort Dougan worked on high speed MPI years ago and I worked on fault-tolerance a very long time ago. All the new work, however, benefits from the operation of a real-time kernel above a server kernel.

Security notes

David Elliot Bell has an interesting essay on the US government and software security – little “inside baseball”, but informative. He cites my critique of MILS. One of Bell’s points, which is briefly mentioned in my note, is that composition is harder than many imagine. Systems A and B may meet some security level individually, but not when connected to each other.

Jim Gray and the existence of working software

Jim Gray disappeared and I hope he will reappear unharmed. I met him once at a Supercomputing show when Cort Dougan and had just been snubbed by the AIX developers at the PowerPC booth who explained that the Linux PowerPC port Cort had just written was beneath their contempt. Gray came up to us and was interested in discussing the work we were doing. Gray was one of the main designers of of the Tandem fault tolerant systems – our competitive target back at Auragen in the 1980s. We lost. After Auragen I eventually went back to graduate school and one day a famous “formal methodist” came by to lecture us about how lousy software was and how we all needed to use his brand of mathematical analysis to improve ourselves. The sage pointed out that without such intellectual tools, limited software engineering as practiced by the ordinary folk could never develop anything like a fault tolerant database that was reliable. I tried to explain that Tandem, and Jim Gray, had done exactly that, but data points were not persuasive to the formal methodist.