Closed GoogleCodeExporter closed 8 years ago
reduce_intervals needs to be dispatched via ADL similarly to
default_decomposition to avoid #including system-specific headers
Original comment by jaredhoberock
on 17 Feb 2012 at 12:06
I think it may be best to eliminate generic uses of reduce_intervals from the
backends. At one point I thought it might be a useful sub-primitive, but I
don't believe we make productive use of that anywhere.
I'll take care of this one.
Original comment by wnbell
on 17 Feb 2012 at 12:22
i've already done it
Original comment by jaredhoberock
on 17 Feb 2012 at 12:24
wait, you said you want to eliminate it entirely? it's all yours then
Original comment by jaredhoberock
on 17 Feb 2012 at 12:26
here is a reproducer:
#include <thrust/device_vector.h>
int main()
{
return 0;
}
jhoberock@jhoberock-dt:~/dev/hg/thrust-tot$ g++ -I. test.cpp
-DTHRUST_DEVICE_SYSTEM=THRUST_DEVICE_SYSTEM_OMP
In file included from ./thrust/system/cuda/detail/arch.inl:24:0,
from ./thrust/system/cuda/detail/arch.h:175,
from ./thrust/system/cuda/detail/detail/launch_closure.h:21,
from ./thrust/system/cuda/detail/reduce_intervals.inl:25,
from ./thrust/system/cuda/detail/reduce_intervals.h:51,
from ./thrust/system/detail/internal/reduce_intervals.inl:24,
from ./thrust/system/detail/internal/reduce_intervals.h:49,
from ./thrust/system/detail/internal/reduce.inl:23,
from ./thrust/system/detail/internal/reduce.h:47,
from ./thrust/system/omp/detail/reduce.inl:19,
from ./thrust/system/omp/detail/reduce.h:52,
from ./thrust/system/omp/detail/algorithm.h:33,
from ./thrust/system/omp/memory.h:410,
from ./thrust/detail/adl_helper.h:42,
from ./thrust/detail/reference.inl:27,
from ./thrust/detail/reference.h:144,
from ./thrust/memory.h:25,
from ./thrust/device_ptr.h:25,
from ./thrust/device_malloc_allocator.h:25,
from ./thrust/device_vector.h:25,
from test.cpp:1:
./thrust/system/cuda/detail/guarded_cuda_runtime_api.h:38:30: fatal error:
cuda_runtime_api.h: No such file or directory
compilation terminated.
Original comment by jaredhoberock
on 17 Feb 2012 at 12:29
nevermind, i'll take it
Original comment by jaredhoberock
on 17 Feb 2012 at 12:32
This issue was closed by revision 729c4da38ce5.
Original comment by jaredhoberock
on 17 Feb 2012 at 2:05
Original issue reported on code.google.com by
jaredhoberock
on 16 Feb 2012 at 7:35