These are the editorial corrections to the ninth draft in reverse order. All are present in the tenth (post-ballot) draft.
- Corrected commenter's names
- Changed rationale of eqv? to be normative
- Improved language stating when inexact numbers are not eqv?
- Cleaned up bibliography to include only cited works, including SRFIs
- Added grammar for case-lambda to tail-recursion section
- Warning about <init>s was incorrectly on letrec instead of letrec*
- Fixed minor editorial nits from a last-minute review
- Updated acknowledgements of commenters
- Rule out xbeef as a character name because it is a hex scalar value
- Restored R5RS results of gen-counter and gen-loser
- Cleaned up definition of "variable definition"
- Cleaned up zero-or-more values continuations language
- Falling off a cond-expand is now undefined behavior (no particular behavior was voted on)
- Minor typos fixed (thanks to Jussi Piitulainen)
- Corrected changes not to imply that character and string comparisons accept 0 or 1 arguments
- Added note about the order of arguments in R6RS vs. R7RS bytevector-copy!
- Restored reference to eqv? in the storage model (as opposed to "operational equivalence")
- Improved allocation discussion to handle empty objects specially
- Changed "as following" to "as follows"
- Cleaned up all cases of "forbidden in strings" to be uniform, removing redundant references
- Made the definitions of character comparisons defer to char->integer to handle non-Unicode
- Fixed quasiquotation and symbol-equality examples
- Removed procedure inequivalence from language changes list
- Squeezed vertical space to save a whole page
- Reverted to R5RS equivalence semantics for procedures (substantive change per WG unanimity)
- Removed "equal," from definition of string<? and friends