Subject: [boost] Warning policy? local variable hides (i.e. shadows) global variable
From: Beman Dawes (bdawes_at_[hidden])
Date: 2015-01-15 08:11:16
GCC and Clang have had a warning for years (-Wshadow): "Warn whenever
a local variable or type declaration shadows another variable,
parameter, type, class member (in C++), or instance variable (in
Objective-C) or whenever a built-in function is shadowed. Note that in
C++, the compiler warns if a local variable shadows an explicit
typedef, but not if it shadows a struct/class/enum."
VC++ 2015 (aka 14.0) Preview has now added a similar warning, C4459.
For example, "c:\boost\modular\develop\boost\lexical_cast\detail\converter_lexical_streams.hpp(429):
warning C4459: declaration of 'n' hides global declaration". The VC++
warning also provides location info so is easy to view both the
declarations involved. In this case, the declaration "bool
operator<<(short n)" is shadowing a variable name in the unnamed
namespace of the calling translation unit.
The process of clearing these shadow warnings occasionally finds bugs
that are otherwise difficult to detect. Peter Dimov and I have both
found bugs in our code in the process of clearing the new C4459
warning. Even though most of the warnings don't actually signify bugs,
I'm finding that clearing them makes code clearer and less confusing,
particularly code I haven't looked at in a long while.
Should Boost have policy to clear these warnings?
A lot of the warnings involve function argument names. Should we have
a guideline to prevent shadow warnings? A convention for argument
names would make it easier to submit pull requests. Possible
* Prefix function argument names with "a_". Rationale: The "m_" prefix
for member names has been a success.
* Suffix function argument names with "_". Rationale: Short and less
distracting than "m_" prefix.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk