Frank Mori Hess:

On Monday 13 April 2009, Beman Dawes wrote:

...

Signals2 is failing five tests on all Mac OS testers.

See http://beta.boost.org/development/tests/release/developer/signals2.html

Please fix or markup.

The failures seem to be due to some bug particular to gcc 4.0 that affects bind. See for example:

http://beta.boost.org/development/tests/release/developer/output/Sandia-leop...

Has anyone seen this before and have a suggestion for a workaround?

I haven't. What seems to be happening is that when you pass _1, which is inline arg<1> _1() { .. } to A1 const& a1, A1 is properly deduced as a function type, but the const is not dropped, as it should be; instead, the type of a1 becomes arg<1> (&)() const (which is illegal BTW). When this a1 is passed (by value) to boost::bind, it decays to its corresponding pointer type, arg<1> (*)() const - which is also illegal. :-) A workaround is to use pass by value instead of A1 const&.

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 13 April 2009, Peter Dimov wrote:

I haven't. What seems to be happening is that when you pass _1, which is

inline arg<1> _1() { .. }

Also, if I modify bind/placeholders.hpp to define _1 as the usual boost::arg<1> _1; the problems seem to go away. Maybe the defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ <= 400) should be modified to additionally check that __GNUC_PATCHLEVEL__ is less than 1? I ran the bind tests and they still all pass with g++ 4.0.1 with this change.

to A1 const& a1, A1 is properly deduced as a function type, but the const is not dropped, as it should be; instead, the type of a1 becomes arg<1> (&)() const (which is illegal BTW). When this a1 is passed (by value) to boost::bind, it decays to its corresponding pointer type, arg<1> (*)() const - which is also illegal. :-)

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEARECAAYFAknjkcIACgkQ5vihyNWuA4WdIACg3MNMqwlrBRI9cJRvM6M0m3zs HkwAoMKf+0HQL+g7nc05AdygbRmrOXCs =IgII -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 13 April 2009, Peter Dimov wrote:

There is no test for this. You need to verify that bind.hpp still works with precompiled headers when included in more than one translation unit.

Oh I see. No, it doesn't work in that case, I get errors like: main.o:(.bss+0x8): multiple definition of `(anonymous namespace)::_1' a.o:(.bss+0x8): first defined here collect2: ld returned 1 exit status However, using the definitions of the placeholders provided for MSVC: static boost::arg<1> _1; does work for gcc 4.0.1, for all cases: the bind tests, multiple translation units with a precompiled header, and forwarding of the placeholder as a const reference. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEARECAAYFAknjpOUACgkQ5vihyNWuA4Wa7wCglMaW6B1AsggUKdToT7GfqPZS aBQAnjbdynw/ZrzPPGjIzbltF7NPfNu8 =wKeh -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 13 April 2009, Peter Dimov wrote:

Frank Mori Hess:

...

However, using the definitions of the placeholders provided for MSVC:

static boost::arg<1> _1;

does work for gcc 4.0.1, for all cases:

It works, but produces annoying "variable not used" warnings. :-)

I haven't seen them, using -Wall and -Wunused-variable. In any case, gcc does have an "unused" attribute which can be used to suppress that warning for specific variables. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEARECAAYFAknjqYAACgkQ5vihyNWuA4W0UACcCruS/cal7Qodifa/FWWB37dJ CfEAn0RaH/kvbwRJV61vUurf+yPvaBV4 =BxY/ -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday 13 April 2009, Peter Dimov wrote:

If you want to switch g++ 4.0 to the static placeholder path, that's fine with me.

Ok, I've done it on trunk with revision 52384. I'll merge it to release in a couple days after some of the tests cycle, unless you tell me not to. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEARECAAYFAknklQEACgkQ5vihyNWuA4VfrACeMQBFaeOMXOGrq9vtrZsIsELU mVIAoMRe0p9/ue9VSFdusJj0LfwX6Nn9 =mGIZ -----END PGP SIGNATURE-----