[mpl] Can I merge pull requests for 'develop' into 'master' for Boost 1.56
Eric Niebler has issued a couple of pull requests for MPL on 'develop' involving VC++ and the has_xxx functionality which I have merged to 'develop' after testing them locally and successfully with various versions of VC++ from VC8 to VC12. The earlier already cycled through regression tests successfully but the latest was merged just recently. The latest involves a change for the NVidia CUDA compiler, for which I see no regression testing. I have not tested this change with the CUDA compiler ( is there even a Boost Build toolset for this compiler ? ) but I implictly trust Eric's judgment. Can/should these changes be merged to 'master' for 1.56 ?
On 07/07/2014 05:30 PM, Edward Diener wrote:
Eric Niebler has issued a couple of pull requests for MPL on 'develop' involving VC++ and the has_xxx functionality which I have merged to 'develop' after testing them locally and successfully with various versions of VC++ from VC8 to VC12. The earlier already cycled through regression tests successfully but the latest was merged just recently. The latest involves a change for the NVidia CUDA compiler, for which I see no regression testing. I have not tested this change with the CUDA compiler ( is there even a Boost Build toolset for this compiler ? ) but I implictly trust Eric's judgment.
Can/should these changes be merged to 'master' for 1.56 ?
Well, I'm a release manager, and I think it's ok. I admit I'm biased and should probably recuse myself. Maybe one of the other release managers could chime in? \e
On 9 July 2014 07:04, Eric Niebler <eniebler@boost.org> wrote:
On 07/07/2014 05:30 PM, Edward Diener wrote:
Eric Niebler has issued a couple of pull requests for MPL on 'develop' involving VC++ and the has_xxx functionality which I have merged to 'develop' after testing them locally and successfully with various versions of VC++ from VC8 to VC12. The earlier already cycled through regression tests successfully but the latest was merged just recently. The latest involves a change for the NVidia CUDA compiler, for which I see no regression testing. I have not tested this change with the CUDA compiler ( is there even a Boost Build toolset for this compiler ? ) but I implictly trust Eric's judgment.
Can/should these changes be merged to 'master' for 1.56 ?
Well, I'm a release manager, and I think it's ok. I admit I'm biased and should probably recuse myself. Maybe one of the other release managers could chime in?
Is CUDA emulating Visual C++? If it is, then you should use _MSC_VER rather than BOOST_MSVC, otherwise it will break again if config ever gets CUDA support. Do you know if it works with CUDA on other platforms?
On 07/09/2014 01:43 AM, Daniel James wrote:
On 9 July 2014 07:04, Eric Niebler <eniebler@boost.org> wrote:
On 07/07/2014 05:30 PM, Edward Diener wrote:
Eric Niebler has issued a couple of pull requests for MPL on 'develop' involving VC++ and the has_xxx functionality which I have merged to 'develop' after testing them locally and successfully with various versions of VC++ from VC8 to VC12. The earlier already cycled through regression tests successfully but the latest was merged just recently. The latest involves a change for the NVidia CUDA compiler, for which I see no regression testing. I have not tested this change with the CUDA compiler ( is there even a Boost Build toolset for this compiler ? ) but I implictly trust Eric's judgment.
Can/should these changes be merged to 'master' for 1.56 ?
Well, I'm a release manager, and I think it's ok. I admit I'm biased and should probably recuse myself. Maybe one of the other release managers could chime in?
Is CUDA emulating Visual C++? If it is, then you should use _MSC_VER rather than BOOST_MSVC, otherwise it will break again if config ever gets CUDA support. Do you know if it works with CUDA on other platforms?
Honestly, I don't know how the CUDA compiler works. I know it *does* work with CUDA on Linux with gcc. I also know this change was needed on Windows.
On 7/9/2014 11:54 AM, Eric Niebler wrote:
On 07/09/2014 01:43 AM, Daniel James wrote:
On 9 July 2014 07:04, Eric Niebler <eniebler@boost.org> wrote:
On 07/07/2014 05:30 PM, Edward Diener wrote:
Eric Niebler has issued a couple of pull requests for MPL on 'develop' involving VC++ and the has_xxx functionality which I have merged to 'develop' after testing them locally and successfully with various versions of VC++ from VC8 to VC12. The earlier already cycled through regression tests successfully but the latest was merged just recently. The latest involves a change for the NVidia CUDA compiler, for which I see no regression testing. I have not tested this change with the CUDA compiler ( is there even a Boost Build toolset for this compiler ? ) but I implictly trust Eric's judgment.
Can/should these changes be merged to 'master' for 1.56 ?
Well, I'm a release manager, and I think it's ok. I admit I'm biased and should probably recuse myself. Maybe one of the other release managers could chime in?
Is CUDA emulating Visual C++? If it is, then you should use _MSC_VER rather than BOOST_MSVC, otherwise it will break again if config ever gets CUDA support. Do you know if it works with CUDA on other platforms?
Honestly, I don't know how the CUDA compiler works. I know it *does* work with CUDA on Linux with gcc. I also know this change was needed on Windows.
I had not merged your first change from 'develop' to 'master' and I am sure that is correct, since I tested it with VC++. I have VC8 through the latest VC12 I test against. Were you able to test your second change and how did you do it, since there is no toolset for CUDA AFAICS ? I assume you figured out a command line to pass to the CUDA compiler if you tested it. If it is alright to merge these changes to 'master' should I tell Marshall Clow after it has been done ?
On 07/09/2014 09:04 AM, Edward Diener wrote:
I had not merged your first change from 'develop' to 'master' and I am sure that is correct, since I tested it with VC++. I have VC8 through the latest VC12 I test against.
Were you able to test your second change and how did you do it, since there is no toolset for CUDA AFAICS ? I assume you figured out a command line to pass to the CUDA compiler if you tested it.
I'm working on a cross-platform project that uses CUDA and Boost. I use CMake to find CUDA and build the .cu files. It generates the command line. It's a black box as far as I'm concerned. Everything worked on Linux and Windows before my first change, broke on Windows after my first change, and works again after my second change.
If it is alright to merge these changes to 'master' should I tell Marshall Clow after it has been done ?
I'm sure it'll get picked up eventually. *Lots* of important changes will be made before the beta and the final release. I'm not worries about this getting dropped. Eric
On 7/9/2014 12:51 PM, Eric Niebler wrote:
On 07/09/2014 09:04 AM, Edward Diener wrote:
I had not merged your first change from 'develop' to 'master' and I am sure that is correct, since I tested it with VC++. I have VC8 through the latest VC12 I test against.
Were you able to test your second change and how did you do it, since there is no toolset for CUDA AFAICS ? I assume you figured out a command line to pass to the CUDA compiler if you tested it.
I'm working on a cross-platform project that uses CUDA and Boost. I use CMake to find CUDA and build the .cu files. It generates the command line. It's a black box as far as I'm concerned. Everything worked on Linux and Windows before my first change, broke on Windows after my first change, and works again after my second change.
If it is alright to merge these changes to 'master' should I tell Marshall Clow after it has been done ?
I'm sure it'll get picked up eventually. *Lots* of important changes will be made before the beta and the final release. I'm not worries about this getting dropped.
I have pushed your changes to mpl 'master' after checking them out once again against VC8 through VC12.
On 9 July 2014 16:54, Eric Niebler <eniebler@boost.org> wrote:
On 07/09/2014 01:43 AM, Daniel James wrote:
Is CUDA emulating Visual C++? If it is, then you should use _MSC_VER rather than BOOST_MSVC, otherwise it will break again if config ever gets CUDA support. Do you know if it works with CUDA on other platforms?
Honestly, I don't know how the CUDA compiler works. I know it *does* work with CUDA on Linux with gcc. I also know this change was needed on Windows.
It looks like it's pretending to be Visual C++. It's currently fooling config as it doesn't have any support, but if it does in the future then BOOST_MSVC will no longer be defined and this will break again, so I think the correct macro check is: || (defined(__CUDACC__) && defined(_MSC_VER)) Or maybe: || (defined(__NVCC__) && defined(_MSC_VER)) Since that's the compiler identifier.
participants (3)
-
Daniel James -
Edward Diener -
Eric Niebler