Experiment request: macro fragment fields

[joshtriplett]

This is a request for a lang-team experiment for macro fragment fields (RFC 3714).

As a lang-team member, I don't need a liaison, but I do need another lang-team member to second this experiment.

I'm planning to substantially rework that RFC, as part of my ongoing work making declarative macros better. In the process, I have some experimental implementation work, which I'd like to land under the auspices of a lang-team experiment. The corresponding feature gates would be marked as incomplete until the RFC is updated and approved.

The scope of this work would be:

• Adding some additional macro fragment types, such as :fn and :adt.
• Adding initial support for macro fragment fields such as ${f.name}.

[scottmcm]

I'm happy to second this. If I can add something to your list, how about a :field_name so things like offset_of have something better than ident-or-literal to avoid the 0.0 problems and to block string literals sooner. And I'm definitely interested in giving people a better way to match definitions instead of needing to attempt to build it all up manually. (I worry about what it means to add things they don't handle to those lists -- imagine what it means to add default field values to the struct parser, for example -- but that's a thing to figure out on the way, not a reason to block an experiment.)

[traviscross]

We reviewed this for context in the lang call today. Thanks @joshtriplett for pushing forward on the experiment.