-----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost- bounces@lists.boost.org] On Behalf Of Ion Gaztañaga Sent: Sunday, October 21, 2007 2:10 PM

Which is(are) the name(s) of the toolset(s)? In the regression tests

http://beta.boost.org/development/tests/trunk/developer/interprocess.ht ml

I can't find any 64 bit windows toolset.

The regression test report generation has been having problems for the last few days and many of the test results are disappeared. See my post to the testing mailing list. The failing toolsets: intel-vc8-win-10.0_x86_64/address-model=64/instruction-set=Nocona msvc-8.0_64/address-model=64/instruction-set=Nocona msvc-8.0_x86_64/address-model=64/instruction-set=Nocona Regards, Sean

-----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost- bounces@lists.boost.org] On Behalf Of Ion Gaztañaga

Marked as unusable for Interprocess. I'll need to install 64 bit Windows version to find out what's happening, but this might take some time...

I'm not sure how this would solve my problem. If I understand you correctly, you only marked errors as expected but this does not prevent tests from running on Win64 systems. Regards, Sean

-----Original Message----- From: boost-bounces@lists.boost.org [mailto:boost- bounces@lists.boost.org] On Behalf Of Ion Gaztañaga

While I manage to get access to a Win64 platform, can you give some hint on which assertions are failing?

Ion, Looks like this following code in mapped_region.hpp (line 237): if(!mhandle.is_shm){ //Update mapping size if the user does not specify it if(size == 0){ __int64 total_size; if(!winapi::get_file_size(detail::file_handle_from_mapping_handle(mappingge t_mapping_handle()), total_size)){ error_info err(winapi::get_last_error()); throw interprocess_exception(err); } if(total_size > (__int64)((std::size_t)(-1))){ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ error_info err(size_error); throw interprocess_exception(err); } size = static_cast<std::size_t>(total_size - offset); } Causes an exception to be thrown. There is an access violation during stack unwinding (unfortunately, I did not have time to track that down, you might want to force an exception in the 32-bit version to see the problem). Regards, Sean

Sean Huang escribió:

if(total_size > (__int64)((std::size_t)(-1))){ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ error_info err(size_error); throw interprocess_exception(err); } size = static_cast<std::size_t>(total_size - offset); }

Sometimes I must admit my code is *really* bad. The ugly castswill make that fail for 64 bit machines (where sizeof(std::size_t) == sizeof(__int64). I've committed a fix:: if(total_size > std::numeric_limits<std::size_t>::max()){ Let's see if things go better. Anyway, I'll continue to chatch the access violation. Regards, Ion

Bronek Kozicki wrote:

Sean Huang <huangsean@hotmail.com> wrote:

...

Interprocess causes numerous "debug assertions" on Windows 64-bit platform. Since nobody is clicking on the dialog boxes and the process will only be killed by bjam after 5mins (necessary for some of the libraries), this results in unbearable long testing times.

some time ago I posted fix designed specificaly to close such windows ...

Which, AFAIR, I applied to bjam. Bronek, perhaps you could check the bjam code to make sure I didn't overlook something. Since I did rewrite the exec code not so long ago. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo