https://bugs.llvm.org/show_bug.cgi?id=41100
Bug ID: 41100
Summary: "in call to f(x)" note should pretty-print original
value of argument 'x', not mutated value
Product: clang
Version: trunk
Hardware: PC
OS: All
Status: NEW
Severity: normal
Priority: P
Component: -New Bugs
Assignee: unassignedclangb...@nondot.org
Reporter: arthur.j.odw...@gmail.com
CC: htmldevelo...@gmail.com, llvm-bugs@lists.llvm.org,
neeil...@live.com, richard-l...@metafoo.co.uk
When Clang prints diagnostic notes of the form "...in call to f(arg1,
arg2)...", it seems to lose track of the arguments' ORIGINAL values. Clang ends
up pretty-printing the CHANGED values of those arguments. This can be very
confusing, especially when there's recursion involved, because the text of the
diagnostic will appear to be talking about a call that never appears in the
source code (or worse, a call that coincidentally DOES appear in the source
code, but isn't the culprit). Here's a reduced test case:
```
// https://godbolt.org/z/PIL0lW
constexpr int f(int i) {
i = -i;
return 1 << i;
}
static_assert(f(1));
```
The problematic part of the following error message is where it mentions "call
to 'f(-1)'". It should mention "call to 'f(1)'" instead.
```
<source>:7:15: error: static_assert expression is not an integral constant
expression
static_assert(f(1));
^~~~
<source>:4:14: note: negative shift count -1
return 1 << i;
^
<source>:7:15: note: in call to 'f(-1)'
static_assert(f(1));
^
1 error generated.
```
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
