Repository navigation
CWG3221 [diff.stat] Add entry for C and C++ defining the scope of for loop identifiers differently #984
Description
Full name of submitter (unless configured in github; will be published with the issue): Jay Ghiron
Reference (section label): [diff.stat]
Issue description:
See C DR 466:
static inline int f (void) {
for (int i = 0; ; ) {
long i = 1; // valid C, invalid C++
// ...
return i; // (perhaps unexpectedly) returns 1 in C
}
}In C2Y the same also applies to if and switch.
Partial suggested resolution:
Create a new paragraph at the end of [diff.stat]:
Affected subclause: [stmt.for]
Change: Names declared inside of the init-statement of a for loop have the same scope as names declared directly in the statement of the same for loop.
[Example 1:for(int i=0;;){ int i=1; // valid C, invalid in C++ return i; }- end example]
Rationale: <to be decided>
Effect on original feature: Change to semantics of well-defined feature.
Difficulty of converting: Syntactic transformation. A second compound statement will allow reusing the name.
How widely used: Seldom.
Activity
t3nsor commentedon Aug 21, 2026
That's not actually correct: [basic.scope.block] says that each iteration statement introduces a block scope, but the substatement also introduces a block scope. So there are two different scopes, but you're not allowed to declare an
iin both of them because we have a special rule prohibiting it.I don't know the exact rationale, and a vague "catch some bugs" seems unconvincing to me. I've suggested in the past that we abolish all special rules of this kind.
Halalaluyafail3 commentedon Aug 22, 2026
Perhaps a better way to describe it would be:
jensmaurer commentedon Aug 23, 2026
CWG3221
Removing the special-case rule is probably an EWG matter.
[-][diff.stat] Add entry for C and C++ defining the scope of for loop identifiers differently[/-][+]CWG3221 [diff.stat] Add entry for C and C++ defining the scope of for loop identifiers differently[/+]Halalaluyafail3 commentedon Aug 27, 2026
I believe the possessive form should be C++'s instead of C++', at least that is what the note in [basic.pre]/1 uses.