backwards headers vs. C++0x

[David Fang]

> > My original suggestion/patch was to remove all, completely. However,
> > I'm  now suggesting to soften this to: removal, then allow old headers
> > to petition for re-instatement. Providing, of course, that we get to
> > abuse the petitioner as part of this process. That way, dead headers
> > will indeed die off. The rest will have to show signs of life.
> >
> > Gaby seems to prefer that we keep most of them. I think this is what
> > Howard prefers as well. Do I have this about right?
> >
> > Paolo? Ulrich? Loren? Others?
>
> I'm not as high up as the names you list, but I'll give my comment anyway :)
>
> I think they should be removed. In particular I believe compilers (and
> their standard libraries) should aim to exactly implement the
> standard. There is no reason I can see that these backward headers
> couldn't just be packaged in a seperate mini-library which people can
> use if they want, and that would have the advantage using them
> wouldn't tie them to g++.

Hi all,
	While I'm in favor of removing the backwards headers outright, I
wouldn't object just hiding them away (remove backwards/ from the default
search path, for instance) and providing an option to ask for them
explicitly.  "Really, this is your LAST chance, one last chance to fix
your includes!"
	Another option is to promote #warning to #error, so at
least users understand what to do (along the lines of the current
message), as opposed to "<iostream.h> header not found".

Fang