[I6] Using before_implicit in 6.34 generates RTEs/ignores pluralname?

Authoring Inform 6
1

I was looking through the release notes for Standard Library 6/11, and I wanted to try out the before_implicit property.

Object grapes "grapes" Start
    with    name 'grapes',
            description
                "They have before_implicit 3.",
            before_implicit 3
    has     edible pluralname;

Using Inform 6.34/Standard Library 6.12.4, when trying eat the grapes while not holding them, the response is:

>EAT GRAPES

[** Programming error: tried to find the "parent" of nothing **]

[** Programming error: tried to test "has" or "hasnt" of nothing. **]
You aren't holding that.

which, in addition to generating RTEs, apparently ignores the pluralname property set for the grapes object. The same occurs using Inform 6.35.

Using Inform 6.31/Standard Library 6/11, the RTEs don’t occur, and the message is slightly different:

>EAT GRAPES
You aren't holding that!

Is there something wrong with the definition of the grapes object?

2

I looked at the code in parser.h (StdLib 6.12.4), and, with respect to the disregard for the pluralname attribute, I think the relevant block (a bit after .GiveError) is:

if (etype == NOTHELD_PE) {
            if (parent(noun) has container)
                    L__M(##Take, 14, noun);
            else
                    L__M(##Miscellany, 32, not_holding);
            oops_from=saved_oops;
}

In english.h, the LanguageLM() routine’s Miscellany: block contains:

    32: CSubjectIsnt(actor); " holding that.";

Should this be:

    32: CSubjectIsnt(actor); " holding ", (ThatOrThose) x1, ".";

instead?

With “messages” on, the RTEs seem to be being generated during a call to LibraryExtensions.RunWhile(), but I haven’t figured out exactly what’s going on.

EDIT: It’s not LibraryExtenstions.RunWhile(), after all. (That’s just the last message passed before the RTEs.) Both RTEs are being generated by the same block, when testing condition (parent(noun) has container).

A modification to check for noun being nothing fixes it. (FYI, @DavidG)

if (etype == NOTHELD_PE) {
	if (noun ~= nothing && parent(noun) has container) ! THIS LINE MODIFIED
		L__M(##Take, 14, noun);
	else
		L__M(##Miscellany, 32, not_holding);
	oops_from=saved_oops;
}