Document why we need unquoted literals

[stasm]

In #477, we're documenting the design of quoted literals. We should have a similar document about unquoted literals. In particular, we have good reasons for unquoted literals to be used as variant keys and option values. There are also use-cases for using them in operand positions, especially number literals.

I think a design doc about the unquotes literals would be useful in the context of the following ongoing discussions:

• How close to nmtoken do we want the unquoted literals to be?
• What syntax and sigils do we want to use for open/close elements?
• If and how to allow negative numbers as unquoted literals?
• If and how to make literals in operands and literals in keys/optvals follow the same syntax rules?
• What syntax to adopt or recommend for namespacing?

[aphillips]

Looking at the question list...

How close to nmtoken do we want the unquoted literals to be?

We have aligned on NCName instead of nmtoken

What syntax and sigils do we want to use for open/close elements?

This is a separate conversation that we are close to resolving. Alignment on NCName actually avoids this being in conflict.

If and how to allow negative numbers as unquoted literals?

We added number-literal to the syntax to permit this. Moving from +/- to hash-and-slash will remove the remaining issue. Of course, we need to agree to hash-and-slash...

If and how to make literals in operands and literals in keys/optvals follow the same syntax rules?

This is done?

What syntax to adopt or recommend for namespacing?

This is done.

Assuming we can reach consensus on spannables, I believe this will be complete. Do you still think we need a design doc to capture the requirement?

[stasm]

You're right, the alignment on NCName helped a lot here.

My original request for a doc was made back when there were more open questions, e.g. about namespacing and syntax conflicts. Since then we've made good progress on many of these questions; others are being actively discussed in their own respective docs. Let's close this issue.