This reminds me a little of baudot, which was 5bit but the shift character was stateful. So you had two character sets (nominally "letters" and "figures"). So sending "AA11" would be five symbols (A, A, FIGS, 1, 1), but sending "A1A1" would be seven (A, FIGS, 1, LTRS, A, FIGS, 1).
It entertains me how many of these issues were addressed long before we had computers.
(Aside; could I argue morse was a variable-length encoding?)
_.___ . ...
Re:Four Column ASCII (2017)
by Stratoscope on Thursday September 26, 2019 @03:36AM
(#21079005)
What Morse Code character is "dah dit dah dah dah"? It's not one I recognize.
Ah, this must be what you meant, yes?
—•—— • •••
-.-.
I think some fonts are not great at rendering Morse code :)
There may be a dah dit dah dah dah, but I don't know it either. But then again I don't know the less common punctuation, Cyrillic Morse, and I think heard once there is even Kanji Morse.
...-.-
Many Compose key sequences are based on this.
So ASCII, in a way, may well be the first -or one of the first- variable-length codeset.
Note that overstriking upper-case letters to add diacritical marks does not work for ASCII (see more below).
Also, overstriking was how one typed diacritical marks back in the days of mechanical typewriters. Spanish used to (and may still? I forget) permit upper-case letters to not carry diacritical marks precisely because it was difficult or costly to get typewriters and printers to print such letters. Adding diacritical marks to upper-case letters requires fonts designed for that, and clearly an overstrike sequence cannot work unless the typewriter/printer holds printing the character until it knows whether the next character is BS.