Re: RFC: change in std::type_info's equality operator

[Matt Austern <austern at apple dot com>]

On Feb 4, 2004, at 7:21 PM, Douglas Gregor wrote:

> On Wednesday 04 February 2004 12:58 pm, Matt Austern wrote:
> > Right now, depending on whether or not a certain macro is set when
> > libstdc++ is built, t1==t2 is equivalent to one of these two
> > expressions:
> >   (1) t1.__name == t2.__name, where __name is the address of the 
> > string
> > representing the type's name.
> >   (2) t1.__name == t2.__name || strcmp(t1.__name, t2.__name) == 0.
> >
> > The library uses version (1) iff the compiler defines __GXX_WEAK__ to
> > be true.
> >
> > My proposal: change the library so it unconditionally uses version 2.
> > That is, always fall back to string comparison if the address
> > comparison
>
> Please do! I've heard many complaints about type_info::operator== not 
> working
> across shared libraries, and we have (or need) workarounds in Boost in
> several different places to deal with this.
>
> This particular quirk always surprised me, because it's one of the few 
> things
> that just doesn't work across shared libraries.

Do you happen to know what systems it fails on, and what the use cases
look like?

The reason I ask: seems to me that if you're on ELF or anything at all
like it (i.e. anything where SUPPORTS_ONE_ONLY is true), then weak
symbols should be merged across shared libraries and it'll just work as
expected.

Hm...  Just occurred to me that maybe there is one obscure system where
SUPPORTS_ONE_ONLY is true but where symbols would only get
merged within a single shared library.  I wouldn't be surprised if MS-
Windows met both of those criteria.  Maybe this really is a case where 
the
compiler and the support library are being too ELF-centric.

			--Matt