Bug 463 – [module] private module members have global bindings instead of local ones

http://www.digitalmars.com/d/attribute.html > Private means that only members of the enclosing class can access > the member, or members and functions in the same module as the > enclosing class. Private members cannot be overridden. Private > module members are equivalent to static declarations in C programs. === d.d === private void foo(){ } ====== dmd -c d.d readelf -s d.o > Symbol table '.symtab' contains 11 entries: > Num: Value Size Type Bind Vis Ndx Name <snip> > 10: 00000000 5 FUNC GLOBAL DEFAULT 12 _D1a3fooFZv === c.c === static void foo(){ } ====== gcc -c c.c readelf -s c.o > Symbol table '.symtab' contains 11 entries: > Num: Value Size Type Bind Vis Ndx Name <snip> > 5: 00000000 6 FUNC LOCAL DEFAULT 1 foo This becomes a problem if a private function with C calling convention is defined: === bug.d === private extern(C) void foo(){ } ====== dmd -c bug.d > Symbol table '.symtab' contains 11 entries: > Num: Value Size Type Bind Vis Ndx Name <snip> > 10: 00000000 5 FUNC GLOBAL DEFAULT 12 foo The last line should have been > 10: 00000000 5 FUNC LOCAL DEFAULT 12 foo

I could be wrong, but I don't think that "equivalent to static" meant it's equivalent in how the two are emitted to the object, but rather where they can be used from (i.e. from within a single module).

In D2, readelf now outputs WEAK for foo in all the examples in the original bug report. WEAK is like global but the symbol can be overriden. Is there a situation where this is a problem? If not, can we close this?

Closing as WORKSFORME.

AIU in C local symbols are used to declare functions with the same name in different modules - private functions. They shouldn't override each other and both should be called by their respective modules. They have a pitfall when a public function is inlined and ends up calling private function from a different module, linker rejects such reference to local symbol from a different module.