Add "Allocation" feature for better link validation

[robertberrington]

An item.allocation could be added to allow the user to list the Child documents where requirements should link.

A WARNING can be thrown if:

• The Parent document is missing a Child item linking to it
• The Child document link points to a Parent whose allocation does not include the child

This would allow a more intelligent WARNING for no links from child document to only list the documents where links should exist.

It seems a simple "space-delimited" field would be sufficient.

Consider an ERROR if the document ID does not exist.

[jacebrowning]

This feature request is mostly clear, but I have a few questions.

An item.allocation could be added to allow the user to list the Child documents where requirements should link.

Is this attribute now required on all all items?

• The Parent document is missing a Child item linking to it

Do you mean "Parent item"?

It seems a simple "space-delimited" field would be sufficient.

I'd probably go with a YAML list for simplicity of parsing. What if this attribute is left blank? Are zero links expected?

[robertberrington]

@jacebrowning

Is this attribute now required on all all items?

Yes, but only normative=TRUE should be populated. Perhaps throw a fault if normative=TRUE and item.allocation is empty. Also, throw a fault if normative=FALSE and item.allocation is NOT empty.

I was thinking really hard if I could come up with an instance where a requirement stops someplace other than at the bottom of the children. That would mean that someone is implementing a design/test on an item that isn't supported by underlying children requirements/implementation/test. I think "flow-down" requirements exist because the child needs to link to the parent, but there's no change.

Do you mean "Parent item"?

Yes.

I'd probably go with a YAML list for simplicity of parsing. What if this attribute is left blank? Are zero links expected?

Sure ... I was thinking that the export to Excel would produce a space-delimited list, and that on import, the space-delimited list would be parsed.

I suppose another thing to consider is if a user decides NOT to use this feature ... do you want to give them option for a --no-alloc-check ... in which case you'd have the current behavior?

[jacebrowning]

Do you want to give them option for a --no-alloc-check?

Yeah, that would be good.