Is it by design that PredicateFactory functions aren't generated for TypedListFieldIndexes? It would be nice to be able to:

filterBucket.PredicateExpression.AddWithAnd(PredicateFactory.CompareValue(ActivityTypedListFieldIndex.GroupId, ComparisonOperator.Equal, groupId))

Or is there an inherent problem with this? Should I be doing this some other way?

Typed lists are build up from entity fields, so you should use the entity field enums instead.

But that requires a knowledge of the way the TypedList is built. I find myself often referring back to the LLBL Project to know what the underlying field is (especially where the field is aliased to another name or the field is a join field and it's not immediately know which table the field is coming from).

I can see that it can be done the other way but it just seems counter-intuitive.

Which fields the typed list is build of is in the comments on the typed list. So if you generate xml documentation from the generated code, you can use that as a guide.

The enums specified in the predicate factory are passed on to the entity field factory. This thus isn't directly doable with typedlist enums.

How can one generate a nice CHM file from generated code (with the style like LLBLGenProReferenceManual.chm) ?

Use NDoc http://ndoc.sourceforge.net