Boost C++ Libraries: Ticket #3485: Boost 1.40 Example.cpp program produces pages of MSVC level 1 warning 4180

Paul A. Bristow — Fri, 25 Sep 2009 08:23:20 GMT

Has been fixed already. Sorry for noise - I was muddled the include order - again :-(

status changed; resolution set

Steven Watanabe — Sun, 27 Sep 2009 02:17:30 GMT

status new → closed
resolution → invalid

status, version, milestone changed; resolution deleted

Paul A. Bristow — Wed, 04 Nov 2009 11:26:34 GMT

status closed → reopened
version Boost 1.40.0 → Boost Release Branch
resolution invalid
milestone Boost 1.41.0 → Boost 1.42.0

I've noticed in passing that at 1_41_1_beta that example.cpp is still producing two pages of warnings for this example.

I think that in lambda_functors.hpp, it needs adding a

#pragma warning(disable:4180) qualifier applied to function type has no meaning;

thus:

#if BOOST_WORKAROUND(BOOST_MSVC, >= 1400) #pragma warning(push) #pragma warning(disable:4512) assignment operator could not be generated #pragma warning(disable:4180) qualifier applied to function type has no meaning; ignored #endif

Sorry for my continued muddle :-(

(Note: Version box does not yet include 1.41)

Steven Watanabe — Wed, 04 Nov 2009 15:36:35 GMT

The example compiles without warnings for me in both the trunk and release branches with msvc 8.0 and 9.0 /W4. I haven't checked the beta yet, although I would expect it to be the same as the release branch.

Steven Watanabe — Wed, 04 Nov 2009 15:47:53 GMT

The beta appears to be okay too, even using the IDE instead of bjam to build.

Paul A. Bristow — Wed, 04 Nov 2009 18:40:35 GMT

Replying to steven_watanabe:

The beta appears to be okay too, even using the IDE instead of bjam to build.

This is bizarre. I have also reported some similar 4180 warnings in boost::math which John could not reproduce despite having a test .cpp program from me.

But I have just tried again enabling MS extensions = no and the warnings disappear.

So disabling MS extensions = Yes (/Za) provokes the warnings!

But John has removing the complained-about const s (from trunk) and my test now compiles free of warnings from boost::math - but with some still from Fusion.

I am unclear if the addition of const to a function is actually prohibited by the C++ Standard, and indeed why const has been added by Joel and John - who are not novices ;-)

So I feel that removing the consts is preferred but disabling the warning would be second best.

Steven Watanabe — Wed, 04 Nov 2009 23:56:05 GMT

Okay. I've duplicated the problem.

status changed; resolution set

Steven Watanabe — Sun, 08 Nov 2009 04:11:22 GMT

status reopened → closed
resolution → fixed

(In [57479]) Deal with warnings with /Za. Fixes #3485