np - thought you might want to know as the current behaviour breaks part of the game.
TigerStripes wrote:Oh, I should also add that using us as a test for your parser is a good choice, given the extent of the material that we have, though I'd certainly recommend you use a few others as well. We do a lot of non-standard things with Inform (which is a plus for your testing), but you'll want to make sure you're testing those areas and features of Inform we don't use. One of the main ones is we don't handle inventory and objects as Inform normally does, nor do we use the spacial placement stuff (x is in the box, y is on the table and so on).
Well I planned to handle these the standard way - containers have a list of objects they contain, and items a contained by property. in/on/part of relationships are just different variations of the same principle. Anything special I need to look out for?
TigerStripes wrote:Please let us know how your development goes.
The tokenizer currently reads through the complete source. This is then grouped in sentences/paragraphs/headings etc. Currently I work on transforming the when play begins sections to get a starting point for an input parser.
Which reminds me of another issue with your source - I had to add a hack for stuff like this "Section 3.1.1". The second dot should be treated as a line end since there is no language construct containing multiple dots that i know of. In fact an inform error message says that no dots at all are allowed in headings because of potential parsing issues. Beats me why you normally do not see this error then...
This error is less critical though since it does not affect the function of the game.