On 06/19/2018 09:30 AM, Mark Regan wrote: > https://www.linuxfoundation.org/blog/security-and-performance-help-mainframes-stand-the-test-of-time/ > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > One obvious error in the article: "...from the introduction of the mainframe in 1964...".
The first practical mainframes were actually constructed in the 1940's, some of the earliest being in Germany. The first "commercial" mainframe in the U.S. was produced in 1949 (sold by Eckert-Mauchly). During the 1950's, there were a number of significant mainframe manufacturers in the U.S. (the biggest being IBM), as well as additional companies in other countries. The author has obviously confused the origin of mainframes with the origin of the specific IBM mainframe architecture (S/360, April 1964) that has evolved into the current IBM z-architecture mainframe. While not the origin of mainframes, this 1964 event was indeed very significant: The concept of separating the architecture design as seen by programs and users from the underlying physical hardware implementation has permitted over 60 years of hardware and architecture enhancements while allowing upward compatibility for much application code. Without that compatibility, corporate reliance on computers would have been severely restricted by the difficulty and expense of migrating applications to new hardware with a different architecture each time the old mainframe was outgrown, wearing out, or no longer supported. I know others in ibm-main are old enough to recall all the architectural differences which could cause grief in migrating programs to new hardware in pre-S/360 days: different character codes, even a different set of supported characters, different data representations, different arithmetic precisions, different bases for numeric data, radical differences in machine language and assembler-level code, radical differences in I/O interfaces, incompatible techniques for controlling job flow. There were invariably some side effects of those differences which became important even for applications written in the high-level languages of the time. Even for those high-level languages that had a formal "standard", compiler implementations for specific machines invariably added some hardware-dependent extensions, and the exact semantics of some statements were allowed to be implementation and hardware dependent in ways that could be significant when migrating to a different machine type. Joel C. Ewing -- Joel C. Ewing, Bentonville, AR jcew...@acm.org ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
