[temp.names] Do not use "template-name" for "name referring to template" #3882
Conversation
|
The wording in this area for template names that are not template-names is ambiguous and there is implementation variance; see core/2019/05/6251 and replies. Note also that the previous paragraph talks only about the template-name case and not the other forms of template names. I don't think it's clear whether the committee intended for P0846R0 to apply to operator-function-ids and literal-operator-ids (or even considered the question). |
zygoloid
force-pushed
the
c++20
branch
3 times, most recently
from
e22901e
to
d4a00d6
Compare
jensmaurer
added
the
needs rebase
|
This has a stray commit adding the foreword. Please rebase + force-push. |
From core/2020/03/8668. The structure of [temp.names]/2 – 3 now (following P0846R0, "ADL and Function Templates that are not Visible"): - Uses _template-name_ basically to handle only the class template cases, and - introduces the notion that names can be considered to refer to templates. However, paragraph 3 is prefaced by "[when] a name is **considered to be a _template-name_** [emphasis mine]", which differs from "when a name is **considered to refer to a template**". In particular, it is a stretch to take a name not having the syntactic form of a _template-name_ (such as an _operator-function-id_) as being considered to be a _template-name_ merely because the name is considered to refer to a template.
hubert-reinterpretcast
force-pushed
the
fix_template_name_name_referring
branch
from
4b694d4
to
3ecd1fd
Compare
Done. |
jensmaurer
removed
the
needs rebase
|
We addressed this in a different way in 84087b3. We now define whether an identifier is grammatically a template-name, rather than whether "a name is considered to refer to a template", so this wording has become meaningful. |
From core/2020/03/8668.
The structure of [temp.names]/2 – 3 now (following P0846R0, "ADL and Function Templates that are not Visible"):
However, paragraph 3 is prefaced by "[when] a name is considered to be a template-name [emphasis mine]", which differs from "when a name is considered to refer to a template". In particular, it is a stretch to take a name not having the syntactic form of a template-name (such as an operator-function-id) as being considered to be a template-name merely because the name is considered to refer to a template.