Re: [RFC] libstdc++/6720 and libstdc++/6671

[Mark Mitchell <mark at codesourcery dot com>]

> The standard does specify the effect of #include <cmath> in a program
> (with on exception, see below): it should make the declarations given
> in section 26 available to the program.
> The specification of all standard headers is given in 17.4.1.2.
>
> The exception is 17.4.3.2:
>
>    *If* a file with a name equivalent to the derived file name for one of
>    the C++ Standard Library headers is *not provided* as part of the
>    implementation, *and* a file with that name is placed in any of the
>    standard places for a source file to be included (16.2), the behavior
>    is undefined.

Thank you for providing that reference.

I agree; we must do some kind of magic.

One possibility is to not provide a file with the right name.  For example,
hardwire the full contents of the header file into the compiler, and then
not provide a header at all.  But, we're certainly not going to go that far.

So, I agree, we need some special casing somewhere to make this work.

If we had a directory that contained *only* the standard headers, i.e,.
<cmath>, <cstdio>, <iostream>, etc., then could we just put it at
the front of the #include path?  In other words, *prepend*
-I /path/to/dir to the front of the users command line?

That would make sure that those header files won.  I guess it would
be better if this affected:

  #include <cmath>

but not:

  #include "cmath"

I'm not sure if cpplib can keep track of separate lists for the two
different syntaxes, but I suppose it could be taught to do so.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com