[diff.library] Consistency for wide char types #3189
Conversation
|
ISO C11 section 7.19 says wchar_t is a declared type (and goes on to talk about other things as macros, so this is hardly an oversight). ISO C11 section 7.28 says char16_t and char32_t are "declared" types. However, ISO C11 section 7.1.3 clearly says that all global identifiers from header files are reserved for macro use. |
|
I think we want "macro or type name" in our updated wording, highlighting both facts. For C++, both the facts that this is not a macro and that it's not a typedef (to some existing type, hampering differentiation in overload resolution) are important. |
jensmaurer
added
the
changes requested
|
Ok! I'll force push that fix to the pull request.
… On 26 Aug 2019, at 22:43, Jens Maurer ***@***.***> wrote:
I think we want "macro or type name" in our updated wording, highlighting both facts. For C++, both the facts that this is not a macro and that it's not a typedef (to some existing type, hampering differentiation in overload resolution) are important.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
|
Pushed changes. |
jensmaurer
added
the
needs rebase
|
Please rebase and fix the conflicts, then force-push. |
|
Rebased. |
jensmaurer
removed
changes requested
|
Hmm, it looks like in the rebase I accidentally pushed just "type names", not "macro or type names". Rebasing and force-pushing. |
[diff.char16] says `char16_t` and `char_32t` > ...do not appear as *macro* names... [diff.wchar.t] says `wchar_t` > ...does not appear as a *type* name... It would be nice to make these consistent if there is no reason for the difference. I thought there might not be because I saw at least one implementation in which `char16_t` and `char_32t` are `typedef`s rather than macros.
[diff.char16] says
char16_tandchar_32t[diff.wchar.t] says
wchar_tIt would be nice to make these consistent if there is no reason for the difference. I thought there might not be because I saw at least one implementation in which
char16_tandchar_32taretypedefs rather than macros.