The stuff having to do with producing an alternate version of the
text is certainly wrong insofar as there's no extra visible copy
produced, but I've always interpreted that language as referring to
the "internal" copy that gets fed to the hash function. It certainly
could be that I've just been around the text and the algorithms long
enough to believe that must be what the current text means so I
didn't think twice about it.
I tried to be careful to distinguish between the language that
describes how the verifier uses l= to manage what's fed to the hash,
which is OK, and the language which suggests that it produces an
edited message, which is not. Take a look and see if you agree.
NOTE WELL: This list operates according to