Parser Puzzle

[mark@kli.org]

OK, here's a strange 'un.  I was running the Lojban KLI info page through
the parser.  Found some mistakes.  (Also found that the parser treats
"kybycy" as a brivla and not a lerfu string, so I changed that to "ky by
cy" in the input but not in the actual file).  But I found a puzzlement.
Here are two sentences.  One, technically, is just a bare vocative.  I
guess since the grammar handles vocatives at the beginnings of paragraphs
specially, I can expect odd behavior, but still.  These two sentences parse
Just Fine separately, but error if there's an {.i} between them.  Since the
first is just a bare vocative the meaning's the same with or without the
{.i}, but it's strange.  Hmm, actually, I think I have managed to bring it
down to a simple case.  It seems to depend on the {ni'o} of all things:

ni'o ju'i loi lobypli .i do cinri

The above is NOT grammatical, according to the parser.  Neither is

ni'o ju'i lobypli .i do cinri

Both ARE correct if the {ni'o} is not there, or if the {.i} is omitted.
Adding {do'u} before the {.i} doesn't help.

This is most puzzling.  I presume the grammar will bear this out when I
check out the EBNF; I further presume this is an oversight, since it
doesn't make much semantic sense to allow it without {ni'o} and forbid it
with one.

What have you folks to say for yourselves?  About this, I mean.

~mark