You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It's confusing that "is constant-initialized" ([expr.const]/2) and "has constant initialization" (([basic.start.static]/2) mean different things.
Moreover, in pretty much every other case of a variable being "meow-initialized" (default/value/copy/direct/list/etc.), the term describes the actual initialization of the variable performed. But in an example like
voidf() {
bool b = std::is_constant_evaluated();
}
b is constant-initialized, but the trial evaluation performed for that determination has nothing to do with how b is actually initialized.
There are only so many ways the words "constant" and "initialize" can be combined, so I don't have a great suggestion. One option might be to limit "constant-initialized" to things where the trial evaluation is consistent with the initialization to be performed - i.e., things that are potentially constant or have static or thread storage duration.
The text was updated successfully, but these errors were encountered:
Editorial meeting feedback: we appreciate the concern. This is not editorial, but also not important enough for Core to explore. Please reopen with a concrete wording suggestion, which we will then run by Core.
It's confusing that "is constant-initialized" ([expr.const]/2) and "has constant initialization" (([basic.start.static]/2) mean different things.
Moreover, in pretty much every other case of a variable being "meow-initialized" (default/value/copy/direct/list/etc.), the term describes the actual initialization of the variable performed. But in an example like
b
is constant-initialized, but the trial evaluation performed for that determination has nothing to do with howb
is actually initialized.There are only so many ways the words "constant" and "initialize" can be combined, so I don't have a great suggestion. One option might be to limit "constant-initialized" to things where the trial evaluation is consistent with the initialization to be performed - i.e., things that are potentially constant or have static or thread storage duration.
The text was updated successfully, but these errors were encountered: