"Parse a C fragment delimited by BEGIN and END.
The BEGIN and END arguments are characters. (Currently, BEGIN can be any
- token type, but you probably shouldn't rely on this.)"
+ token type, but you probably shouldn't rely on this.)"
;; This is decidedly nasty. The basic problem is that `scan-c-fragment'
;; works at the character level rather than at the lexical level, and if we