Sorry, my post was a bit short and unclear. What I meant is that CESU-8 is a **non-standard** form that may be used for internal representations but should not be used in external communication (according to the Wikipedia page on [CESU-8](https://en.wikipedia.org/wiki/CESU-8), which also states that HTML expressly disallows CESU-8). So Tcl should convert to a **standard** encoding for external communication and use what ever it pleases internally. For me, modified UTF-8 seems much closer to the standard (because it differs only in the non-canonical encoding of 0x00) and I felt that it would be better for Tcl to do things the other way round. But of course it's just as non-standard as the other format.