On 14 January 2014 02:26, Paul Gilmartin <[email protected]> wrote: >> On 1/13/2014 7:05 PM, Steve Smith wrote: >>> The main difference among the various flavors of EBCDIC is where the square >>> brackets are. The rest is mostly accented letters. Who knows how that >>> came about, but I'm sure it's not interesting.
Well *I* think it's interesting. But I find lots of things historically interesting that most people seem not to care about. It's more than mere trivia; it's fascinating to find out how we got where we are, whether in geopolitics, technology, or anything else. > Not so much uninteresting as appalling. Various factions at various > times decided that things such as accented characters were more > important than mathematical symbols. Or vice versa. So they > stole the characters they didn't need and replaced them with > those they needed. And so on. Another Whac-a-Mole variant. Well I know how *some* of this came about. Way back when, before the phrase "code page" was in common use, there were 3270s. The early base 3270 (3277) was, of course, a US-English (and initially an upper case only) device, but as soon as the first one was shipped to another country there was a requirement that various country-specific additional characters be supported. These mappings were developed quite ad-hoc, presumably by having some local specialist decide on the characters wanted, paring down the list because of limitations in the amount of then-expensive ROM available, and choosing a mapping of byte values to the new characters, often to match some existing mapping in use for the country-specific print trains or line-mode terminals. When the second generation of 3270s (3278 et al) came out, in 1977 IIRC, ROM was cheaper, and indeed even the US versions could display a fairly large character repertoire - tantalizingly visible by just flipping the terminal switch into TEST mode. Somewhere in here the notion of Country Extended Code Pages (CECPs) came into play. These provided mappings, based on the existing 3277 country-local ones for compatibility, of all the 190 displayable ROM characters, plus the remaining 66 byte values. (The ROM actually had a few more characters, like the clock halves and the stick figure, but these were not program addressible.) Of course not all countries used the same 190 characters, and there were certainly other ROMs available, but for the western European countries the same hardware could be used with just software/microcode customization. Doubtless there were other contributions to the particular 190 character set - from the ASCII world of the day, and also from the various users of 3270s for things like data entry (those superscript numbers 1,2,3, and a couple of others) but that's roughly how we got CECPs - a mix of mapping history from the older 3270s and the need to keep the 66 control characters required by the 3270 architecture. I think almost all of this happened before worldwide standards processes had enough effect on things to make much difference. Prominent vendors other than IBM (e.g. DEC, HP) had their own arbitrary and mostly incompatible ASCII-based character mappings, so it's not all IBM's fault. > Have I ever mentioned that I hate EBCDIC!? Perhaps once or twice. Though why you don't hate ASCII at least as much, I don't understand. Tony H.
