In what way are PC keyboards necessarily limited to 3 levels? I can easily construct a PC keyboard layout using MSKLC in which characters are assigned to Shift+AltGr keystrokes. In fact, the standard US-International keyboard comes like this.
[Alain] I was mainly talking about labeling of keys, as the main difference between commercial keyboards is [so far�] labeling (and the only difference between any national European or Canadian keyboard -- the American keyboard has one key less but even under the hood the electrical contact is there for the extra key, it is just never triggered because over it there is a larger key that bridges the gap).
[Doug]
Why does a 3-shift-state keyboard count as 1 group of 3 levels, but a 4-shift-state keyboard counts as 2 groups of 2 levels? What is the difference, other than the fact that ISO 9995 says there can only be 3 levels?
[Alain] A 4-shift-state keyboard could be 2 groups of two levels (I was implicitly refering to Mark's case where he was talking of 2 languages each using its own alphabet, each one with letter pairs [Latin, Cyrillic, Greek, are perhaps the typical cases he was refering to]), or one group with 3 levels and one group with only 1 level (which would just be quite imbalanced).
There is no restriction at all in the definitions of groups, it just makes sense to build them in a unified way, with some category belonging (but you could also decide that categories are just randomly defined, it's up to the implementer's imagination).
Alain LaBont� Qu�bec
� it would be nice if the computer could have a hint of what is engraved on keyboards, as indicated in skeleton ISO/IEC DTR 15440 (on future keyboards) under ballot, but this would require that the keyboards be queried for this and that they be giving a number back corresponding to a very precise layout without any option (not even a key position changed). So far, PC keyboards just send "scan codes" to the "computer" (regardless of what is engraved, regardless of key positions: this does not help software makers producing a nice user interface: just think about the diagram of the keyboard on a screen which might *not correspond* to the physical beyboard layout)
