table as part of a thing?

Authoring Inform 7
1

Is it correct that a table cannot be part of a thing – as opposed to a list, which definitely can? For instance, this doesn’t compile:

Test is a room. The frob is a thing in Test. The frob has a table called Table of Baz.

Table of Baz
idx	txt
1	"txt 1"
2	"txt 2"

This however compiles fine:

Test is a room. The frob is a thing in Test. The frob has a list of texts called text-lists. Text-lists is usually {"txt 1","txt 2"}.

Why would one data structure (the list) be acceptable to add to an object but not another data structure (the table)?

The reason I ask is that I’ve got a use case where a number of objects each needs to associate with a number of different tables. That association can be managed quite well by creating a list of table names for each thing, but this doesn’t appear to scale well when the list reaches about 30 elements (i.e. table names). There is a noticeable albeit short (<1 second) delay in the response compared to other actions. Presumably this relates to the fact that dynamic lists traverse more slowly than static tables. If each thing could have a table of table names instead of a list of table names, I think all would be sweetness and light.

Is there something I’m missing here? Is there a different/better work-around? Something involving I6 perhaps? TIA.

2

The word you’re looking for is ‘table name.’ (I don’t know why it was necessary to call it something other than ‘table’, while this isn’t necessary with lists - but tables are odd things generally.)

A thing has a table name called current reference. The current reference of a thing is usually Table of Delightful Things. The current reference of frob is Table of Stinky Things.

Edit - I see you’re already using ‘table name’ in other contexts, so I’m not quite sure what it is you’re after. Is the deal that you want to auto-generate tables in the same way that you can auto-generate lists or things and the like?

[code]A nose is part of every person.

A person has a list of texts called hilarious retorts.[/code]
If that’s the issue, no, you can’t do that.

3

Aha, brilliant, and many thanks! The point of this is to get a bit meta and generate responses procedurally from state and data tables stored with each object (which may be a misguided pursuit, but that’s a different matter). Here is a sketch that works just as desired and kind of shows what I’m trying to do (for better or worse):

Test is a room. The frob is a thing in Test. The frob has a table name called ref-table. Ref-table is usually Table of Foo. The frob has a number called state. State is usually 1.

Table of Foo
table name
Table of First
Table of Second
Table of Third

Table of First
idx	txt
1	"T1 txt 1"
2	"T1 txt 2"

Table of Second
idx	txt
1	"T2 txt 1"
2	"T2 txt 2"

Table of Third
idx	txt
1	"T3 txt 1"
2	"T3 txt 2"

Instead of examining frob:
	let id be the state of frob;
	repeat through ref-table of frob:
		let tn be table name entry;
		choose the row with idx of id from tn;
		say "[txt entry][line break]";
	if state of frob is 1:
		now the state of frob is 2;
	else:
		now the state of frob is 1;
	
test me with "x frob/x frob/x frob";

[
expected output:
	
>test me
(Testing.)

>[1] x frob
 T1 txt 1
 T2 txt 1
 T3 txt 1

>[2] x frob
 T1 txt 2
 T2 txt 2
 T3 txt 2

>[3] x frob
 T1 txt 1
 T2 txt 1
 T3 txt 1
]