(This has lots of conflicts with Davis' paper, so this needs to wait a little.)

@tkoeppe , do you want to similarly replace in "typedef name for linkage purposes", which is a stand-alone defined term?

Is "typedef name for linkage purposes" actually defined? I couldn't find the definition, just circular references. (It's also not indexed.)

It looks like "{general} name for linkage purposes" is the defined term, and then "typedef name" is just an application of that -- in which case it suffers form the same problem as reported here. (If instead the whole phrase were a defined term, I'd be happy to keep it and use it.)

dcl.typedef p4

If the typedef declaration defines an unnamed class or enumeration, the first typedef-name declared by the declaration to be that type is used to denote the type for linkage purposes only (6.6).

Is very close to a definition.

Hm, that just sounds like a definition that fills in "typedef-name" into "{general} name for linkage purposes", so that would suggest that we should indeed say "typedef-name for linkage purposes" everywhere. Otherwise we should add something to that paragraph like, "which is called the typedef name for linkage purposes [index]" (but that seems uncalled for).

I've made a suggestion for a proper definition in the linked pull request.