The Mac is like a modern day Betamax 1732

Charlie Gibbs

sorry, i meant to say 'calendar started in 1900'.

i don't know about that, at least the 'working' part. as for mainframes, i remember surprising a few people in the 1980's with the question - "how many mainframes do IBM Amdahl Hitachi etc ship each year?" i recall the answer was somewhere in the hundreds, many of which were replacements.

actually the system was a batch one, originated on 360's in the 1960's, mostly on the mainframe DOS, i believe. used lots of BAL macros. there were consultants who made good money at the time adapting it for lots of businesses - various kinds of record keeping, inventory and so forth, ISAM master files. if you could maintain-extend one, you could handle them all. forget what it was called and where it was invented, perhaps IBM field people. not very interesting today i guess, except to remind me how far things have come.

when customers wanted transaction timestamps for foreign keys in systems that needed to run on multiple platforms (mf-unix-nt) i used to suggest this long character key that had year,month,day,hh,mm,ss and 4 low order digits for milliseconds. i seem to recall that POSIX guaranteed clock granularity of 10 milliseconds but the mainframes were about 1 millisecond. never worried about msdos systems, though. in busy systems (dozens of transactions per second), this seemed to minimize 'convoys' or race conditions because the transaction code wouldn't try more than once to lock a key, it would just generate a new one some milliseconds later and then try again and there was little chance that two transactions would generate the same number. it was OK for the biggest system i saw (48 million telco customers), but i guess there are bigger ones.

pc