Article 163061 of comp.os.vms: In article <32DF9B86.3526@sensor.com>, Ron Natalie writes: >Robert Schuldenfrei wrote: >> In defense of the 6600, it was at a time when it was not at all clear what an OS >> should do and what an "application" should do. NOS was very sparce. As I >> remember every command just executed a program. Even the definition of data >> executed something. > >Yep, you have to realize that the thing was pretty much a batch engine. >All the NOS ugliness got regurgitated into COS on the Crays, familiar >enough, but different enough, to unnerve programmers making the switch. Ah, CDC... You never forget your first computer... I still have a few kilos of CDC core memory (few kilos refers both to the capacity and weight of the unit :-) from a 6400 dismantled long time ago. The OS had some amazing properties you don't find anywhere now. For example all the queues (including executing jobs) survived a deadstart (or reboot as we call it today). Our computer operators had a lot of fun when our local CDC hacker has asked himself a question 'what will happen if my program requests more memory than physically available?' (our CDC had only 192 kwords of memory out of possible 256). So he conducted an experiment - wrote a small program that, when executed, requested 200 kwords of core from the OS. His program got immediately swapped out an set to 'W-MEMRY' state. After a while the author concluded that his experiment was successful, and wanted to kill his job. No way - in order to be killed the job had to be in memory... No problem he thought, a deadstart will kill it... no, it won't. So it seemed for a while that we'll have to live with that thing forever sitting there (or that we'll have to borrow more memory somewhere) - finally on one quiet weekend, when all queues emptied ,the operators restored the system from some old tape and got rid of the monster. Even on VMS swapped-out programs don't survive a reboot (not to mention executing ones). Oh well, maybe it's better so :-) Michal