Potential memory leak? about algorithm HOT 5 CLOSED

I'm not seeing any way to assign to m_Storage.m_dynSet w/o having use_fixed_storage(m_Size) be false. Are you?

This is not how I would write this code today, but I don't see a bug there.

from algorithm.

That may indeed be true. Maybe this is actually a clang-tidy bug. See this conflicting set of logic:

/root/misc/root/smtk-deps/include/boost/algorithm/string/detail/classification.hpp:114:21: note: Taking false branch
                    if(use_fixed_storage(m_Size))
                    ^

and then later, assuming the inverse of it:

/root/misc/root/smtk-deps/include/boost/algorithm/string/finder.hpp:219:55: note: Calling '~is_any_ofF'
/root/misc/root/smtk-deps/include/boost/algorithm/string/detail/classification.hpp:135:51: note: Left side of '&&' is false
                    if(!use_fixed_storage(m_Size) && m_Storage.m_dynSet!=0)
                                                  ^

It seems there are 2 report instances in the description; I'll trim it to just one.

from algorithm.

You could test out your theory by inverting the test at classification.hpp:114 (just change the test to if(!use_fixed_storage(m_Size)) and swap the if and else code blocks) and see if clang-tidy still complains.

from algorithm.

Filed upstream: https://bugs.llvm.org/show_bug.cgi?id=44247

from algorithm.

Closing this, since it's not our bug.

from algorithm.