Keyboard Input

One of the layers from the OS to Squeak level code has to create a character in the extended character set from keyboard input. Because the Windows or Mac sends a multi-octet character code to the Squeak VM, one would imagine that modifying VM to pass multi-octet characters to the Squeak level would be a good solution. However, this is not desirable in two reasons. One is that making a different VM whose difference is merely the keyboard input handling is not worthwhile. The other is that the encoding of the multi octet character sent from the OS is different from one platform to another. It might even differ with the setting of the same OS. For Japanese, for instance, the character sent from the OS could be Shift-JIS on Mac and Windows, EUC-jp or UTF-8 on Unices with different locale settings, or UTF-16 on Windows CE. If we do the translation in the VM, we need to write the C code for each different platform.

In the current implementation of m17n Squeak, we decide not to modify the VM. If the OS sends a multi octet character to the Squeak VM, the VM treats the octets as if they are separable characters and sent them to the Squeak level code through the Sensor object. The Sensor then interprets the characters if necessary and generate appropriate character.