Yes I struggled with whether or not to elaborate on both those points, but I was on my tablet in a motel with crappy wifi so I didn't bother. Quite right on all counts. Whatever. The point is really that zVM doesn't support Unicode so you're stuck with having to translate back to a specific EBCDIC codepage and how you pick which one is appropriate is somewhat of a mystery (to me anyway) since the charset=utf8 doesn't really give you any clues about what kind of codepoints are represented therein. In the domain of platforms that actually support Unicode (which are typically domains that supported ascii and not ebcdic), it doesn't matter, but if you have to go to ebcdic it really does (and there's nothing in the email standards that provides any clues to assist). Internally take a swag at it by looking at the subdomain the mail is directed to (e.g. if it is to a mainframe sited in the USA, we use a US-friendly choice, to one sited in Japan a Japan friendly choice, etc.). We rarely get any complaints that we guessed wrong, but it could easily be because nobody's actually paying attention rather than we're such good guessers. -- bc -- bc
On Sat, Nov 24, 2012 at 5:13 PM, Glenn Knickerbocker <[email protected]>wrote: > On Sat, 24 Nov 2012 12:09:58 -0700, gil wrote: > >Isn't "ascii unicode" somewhat oxymoronic? > > Well, ASCII *is* Unicode--but if what he were translating were ASCII, it > wouldn't need encoding and decoding, since it maps to itself in UTF-8. > > And 8859-1 is Unicode, too, which is why it's easy to decode from UTF-16, > because the UTF-16 encoding just adds a zero byte in front of every 8-bit > character. > > http://users.bestweb.net/~notr/arkville.html "I felt like I was in a > ¬R demented Wallace Stevens poem, with food poisoning." Spalding Gray >
