[llvm-bugs] [Bug 49840] New: Support for branch attribute macros

via llvm-bugs llvm-bugs at lists.llvm.org
Mon Apr 5 02:08:18 PDT 2021


https://bugs.llvm.org/show_bug.cgi?id=49840

            Bug ID: 49840
           Summary: Support for branch attribute macros
           Product: clang
           Version: trunk
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Formatter
          Assignee: unassignedclangbugs at nondot.org
          Reporter: N.James93 at hotmail.co.uk
                CC: djasper at google.com, klimek at google.com,
                    llvm-bugs at lists.llvm.org

I'm not sure how this is best implemented but I couldn't find a nice way to
format code with the c++20 [[likely]] and [[unlikely]] branch attributes if
those attributes are in macro form.

Spelling the attributes normally results in code like this:

if (StartIndex == End) [[unlikely]]
    return npos;

Or if the line needs to be wrapped:

if (StartIndex == End)
    [[unlikely]]
    return npos;

In both these cases its obvious the `return npos;` is the true branch of the if
statement. However when using a macro for these attributes this is no longer
apparent, instead it appears that `BRANCH_UNLIKELY` is the true branch, and
`return npos;` if a statement following the if.

if (StartIndex == End)
    BRANCH_UNLIKELY
return npos;

I set AttributeMacros accordingly, but it seems clang-format doesn't pay
attention to that style setting for if/else branch attributes.

I haven't tested how this plays with switch cases which also support the same
annotations.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20210405/00f2f820/attachment-0001.html>


More information about the llvm-bugs mailing list