Gentle persons: Since there is interest in gathering information for a new Wikipedia article on EMC and since there has been some gratuitous O/S bashing going on recently, I thought I would kill two birds with one stone by recommending y'all read (or reread) the NIST report "Enhanced Machine Control Architecture Overview" NISTIR-5331, by Fred Proctor and John Michaloski [ http://www.nist.gov/manuscript-publication-search.cfm?pub_id=820482 ].*
It nicely describes the intentions of the Systems Integration Group in developing EMC and tells how it fits into the larger Real-time Control System (RCS) architecture championed by Jim Albus. A fundamental goal of their group, like that of my own, was to develop and help promulgate standards that enabled the interoperability of many different, proprietary systems. The EMC work did not pick an operating system, it helped the industry refine functional requirements for operating systems used in machine control. Look, for example, at section 9.1 in the NISTIR. "The operating system is specified by the SOSAS [Specification for an Open System Architecture Standard; from the Next Generation Controller project] to be POSIX-compliant. POSIX does not specific a particular operating system, but the interfaces to systems programmers that operating systems must provide.... "It is anticipated that full support of POSIX functionality (let alone exact compliance) will not be required for a machine controller, but that a subset is more appropriate. Part of the purpose of the implementation of the EMC project, at least as far as the Laboratory Development Controller is concerned, is to define that subset. At this point, it is clear that at a minimum, we need the functionality of the POSIX real-time extensions for semaphores, shared memory, clocks and timers, interprocess communication..., and threads. However, we feel that it is inappropriate and unnecessary to declare that the operating system be POSIX-compliant, in the same way that we would not declare that controllers use the VME bus. [good thing, anyone here still using any VME boards?] Other suitable and popular operating systems exist that we feel vendors will select based on their market." I know Fred and his colleagues struggled to make EMC run on as many different hardware and software systems as they could, precisely because these systems were representative of the market they were serving. As researchers, however, they were limited in what they could accomplish. Had New Mexico Institute of Mining and Technology not been developing RT-Linux in the same time period, Linux could not have been used. Had a similar product for Windows NT been available to the project, Windows NT would have been included in the mix. (After all, even MSDOS was being used in some of the AMRF at the time!) If you look at the original EMC and RCS code, you'll see conditionals for all manner of systems. If I recall, they played with VxWorks for example, which was very popular in some other NIST projects of the time. I'm sure there's conditionals for SunOS and possibly for VAX/VMS. At the time, I had a DEC Alpha and several SGI boxes in my own lab at the other end of NIST. Had I been involved with EMC, we'd probably have some DEC OSF/1 and SGI Irix conditionals in the code too. The bottom line is, don't bash the systems, demand the functionality. Just my 2 cents worth. Regards, Kent *note that the NIST publications are not subject to copyright. While this means you can freely distribute this report, please, as a courtesy to NIST, do not embed it on a website or blog. Rather, post the link for others to access so NIST can show the bean counters downtown that people are accessing its work products, e.g., that its work is relevant. ------------------------------------------------------------------------------ Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing. http://p.sf.net/sfu/novell-sfdev2dev _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
