[LLVMbugs] [Bug 7000] New: Diagnostic for Invalid Cast should give Actual Type Name in addition to Typedef Type Name

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=7000

           Summary: Diagnostic for Invalid Cast should give Actual Type
                    Name in addition to Typedef Type Name
           Product: clang
           Version: 2.7
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: -New Bugs
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: me22.ca at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=4785)
 --> (http://llvm.org/bugs/attachment.cgi?id=4785)
Source file (as pasted in description)

Compiling this bit of code:

    typedef unsigned char value_type;
    template <typename T>
    void update_n(T *p_) {
        value_type const *p = static_cast<value_type const *>(p_);
    }
    int main() {
        char *s;
        update_n(s);
    }

I get the following error from clang++:

    clang.cc:4:27: error: static_cast from 'char *' to 'value_type const *' is
not
          allowed
        value_type const *p = static_cast<value_type const *>(p_);
                              ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    clang.cc:8:5: note: in instantiation of function template specialization
          'update_n<char>' requested here
        update_n(s);
        ^

While it's nice to see the clean name sometimes, in this case knowing about
"value_type" is useless in figuring out why the cast is invalid -- particularly
since it just reiterates what's clear in the caret diagnostic.

In the actual situation from which I took this test case, the value_type
typedef is something derived through a metafunction from template parameters,
so I can't just look at its definition to tell what the type is, so it'd be
much more helpful to know the expanded type.  I'm sure there are plenty of
others where the typedef would be more helpful, and even this time it's nice to
know the typedef name, so it'd be great to have both.  In fact, probably the
nicest I could hope for would be something like this: "static_cast from 'char*'
(T*) to 'unsigned char const*' (value_type const *) is not allowed".  For
implicit casts it might be better to give the typedef names as primary and the
expanded names as parenthetical, though.

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.