Tcl Source Code

Unicode 1.1 left open the possibility that any codepoint in UCS-2 might
one day be assigned. Tcl 8.1 imposes no conditions on the encoding of any
**Tcl_UniChar** value at all.

The standards specifying text encodings publish in the mid-1990s were quite
clear and explicit about the right way to do things. They were often less
demanding and specific about how to respond in the presence of errors. The
spirit of Postel's Robustness Principle

>	*Be liberal in what you accept, and conservative in what you send.*

held considerable influence at the time. Many implementations chose to
accommodate input errors, especially when that was the natural result
of laziness. For example, the specification of FSS-UTF and all specifications
for UTF-8 are completely clear and explicit that the proper encoding
of the codepoint **U+0000** is the byte value **0x00**, (our old friend
the **NUL** byte!).




decoding and strictness

UTF-16 and surrogate pairs