Terms "template declaration" and "template entity" are undefined

[xmh0511]

[temp.pre] p3 states

A template-declaration is a declaration. A declaration introduced by a template declaration of a variable is a variable template. A variable template at class scope is a static data member template.

we have never defined the term template declaration, and the wording template declaration is used more than one time in clause [temp]. As well, the template also depends on the definition of template declaration.

An entity is templated if it is

• a template

As far as now, we only define the variable template. Thus, we may augment the definition for template, as well as, clarify the definition of template declaration.

Change [temp.pre] p3 to that

A template-declaration is a declaration, as known as a template declaration. A template-declaration introduces a template of the entity E(if any) introduced by the declaration in it. A variable template at class scope is shall be a static data member template.

The proposal is fine to clarify a simple explicit specialization is not a template, conversely, a partial specialization is a template declaration, and any restriction applying to a templated entity can also apply to the entity defined in it:

template<class T>
struct A;

template<>
struct A<int>{};  // it is not a template declaration


template<class T>
struct B{};

template<class T>
struct B<T*>{
   void fun() requires (sizeof(T)> 1);  // [dcl.dcl.dcl.decl.general] p4 allows it.
};

Compare to defining a template-declaration is a declaration, I prefer to augment [syntax] to be that

X-declaration is a declaration.

That would cover: parameter-declaration, exception-declaration, and so on.

[frederick-vs-ja]

The term "templated entity" is recursively defined. The crux is possibly that we haven't properly defined the term template, although it's implied that "a template is an entity declared by template-declaration". However, this isn't the proper definition, since a function template can be abbreviated or introduced within a generic lambda.

CWG2862 is related but hasn't addressed this.

Once "template" gets properly defined, I don't think we need to define "X declaration", because it can be parsed as "declaration of X" in plain English.