For supporters, making the hot spots random can be a problem given the different sizes of objects that can be placed on the supporter. So you have to be sure that the spots are chosen so that there are no instances where objects overlap each other visually when placed on the supporter. I'll go into that in more detail later.

As for the parser, I actually did mimic the way Inform 6 does its parsing, so the way you describe the verbs and their syntax structures is exactly the way I've implemented things. The problem is not in the syntax, though, but in the way the parsing is performed. This is very low-level action that is very complicated and difficult to interpret from the Inform code, so for much of it I had to come up with my own interpretation.

The way I've implemented the parser, it is difficult to parse out two nouns that are not separated by a preposition. Mostly that's the result of me coding the parser without putting a lot of forethought into those (relatively less common) situations where two nouns are right next to each other. So I just need to go back and modify the parser so it can do that, which hopefully will not mess up everything that the parser is currently doing well.

I can't remember how many months you take on Vespers, or I can't remember how many time has pass from the release of the pilot of Frayed Knights, but for me, for a graphical 3D ambitious game, from 1 to 5 years of development is quite normal/desirable. More than 5 years… tsk, tsk, it is more kind of waste of time, or lack of time, or lack of resources. (Blue Lacuna was build on 5 years , man, that it is time, for a text game).

Talking about the bed as a supporter… I'm thinking that it could be a good idea to put the location of the hot spots as random. This could give you a more natural feeling of the objects lying on the bed, because now they feel like making a parade.

Talking about the parser and the "reverse actions". Why don't you copy the Inform 6 model, or TADS3 one? If I remember well in the I6 grammar you could specify a kind of grammar as the reverse of an action, so you could say something like this:

!Better than that! Here is the original code for the I6 grammar for the verb Give:

Verb 'give' 'feed' 'offer' 'pay'
* held 'to' creature -> Give
* creature held -> Give reverse
* 'over' held 'to' creature -> Give;

As you can see the inform parser is potent enough to understand that in the phrase:

>give Drogo the keys

"Drogo" is an object, so there's no need of further association with the following words, so it can look up in the next token for the other objects "the keys" (I think parsers usually removes already matched words and continue parsing the next tokens). What I'm trying to explain is… if Inform does it, you can! Just copy it! Copy the greatest, don't reinvent the wheel. You have plenty of design documents about parsers for I6, for I7 and for TADS. And you have the source code as well, so what are you waiting for?

Note the "reverse" attribute in the grammar, that means that the action is programmed to work for

GIVE object TO pnj

So with the phrase GIVE pnj object we only need to reverse the behaviour. That it is the idea, one kind of action, different grammars to say it.

Hope this helps.

Regards, Ruber (aka Urbatain)