On Tue, Jun 7, 2011 at 12:50 PM, t m <span dir="ltr"><<a href="mailto:cmake@majchrowski.co.uk">cmake@majchrowski.co.uk</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Hi<br>
<br>
I've two topics related to the support for custom lang in cmake. I<br>
hope you can help me.<br>
<br>
1) Does anyone knows plan about merge the following feature related to<br>
the CUDA: <a href="http://public.kitware.com/Bug/view.php?id=11887" target="_blank">http://public.kitware.com/Bug/view.php?id=11887</a><br>
<br>
I'm in the process of integrating the CUDA Toolkit into our build<br>
environment and I would need to make a decision which way to go very<br>
soon. I must say that I like much more idea presented in this ticket<br>
instead of using wrapping macros (aka FindCUDA.cmake).I think it's<br>
more elegant and I hope it will be merge to the main stream very soon.<br>
I just need to know which direction will be choose as I don't wont to<br>
be far from the main stream.<br>
<br></blockquote><div><br>Other than the 'more elegant' prospect of being able to just put .cu files into a target, is there anything else deficient about the current FindCUDA macros? All you have to do really is to use cuda_add_executable instead of add_executable.<br>
<br>I've iterated with Peter many times about his proposed changes, and one of the issues we were hoping to solve is the ability to target different output formats. CUDA supports generation of not only object files, but PTX and CUBIN files which need to not be treated as an object file. In addition, it wasn't clear how this adaptation would behave along side the existing FindCUDA behavior for which we have to be backward compatible with.<br>
<br></div><div>James<br><br> <br></div></div><br>