Closed serge-sans-paille closed 1 year ago
@jeanlaroche this patchset brings significant changes, could you give it a try on your codebase?
I will be not before Sunday. Don't have my machine with me...
On 3/31/2023 11:57 PM, serge-sans-paille wrote:
@jeanlaroche https://github.com/jeanlaroche this patchset brings significant changes, could you give it a try on your codebase?
— Reply to this email directly, view it on GitHub https://github.com/serge-sans-paille/pythran/pull/2091#issuecomment-1492855405, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGUBEPPGKK33LDGROJLJ3SDW67GV3ANCNFSM6AAAAAAV7PJH3Y. You are receiving this because you were mentioned.Message ID: @.***>
I'm getting this error in one of my scripts. Do you need me to try to determine where exactly this comes from or can you intuit what the issue is?
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/utils/broadcast_copy.hpp:268:23: error: no member named 'buffer' in
'(anonymous namespace)::pythonic::types::list<long>'
std::copy(other.buffer, other.buffer + other.flat_size(), self.buffer);
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/core.hpp:46:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/slice.hpp:5:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/NoneType.hpp:6:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/builtins/bool_.hpp:6:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/tuple.hpp:13:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/ndarray.hpp:12:
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/utils/broadcast_copy.hpp:268:23: error: no member named 'buffer' in
'(anonymous namespace)::pythonic::types::list<long>'
std::copy(other.buffer, other.buffer + other.flat_size(), self.buffer);
~~~~~ ^
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/utils/broadcast_copy.hpp:300:12: note: in instantiation of function template specialization
'(anonymous namespace)::pythonic::utils::broadcast_copy_helper<(anonymous namespace)::pythonic::types::numpy_gexpr<(anonymous namespace)::pythonic::types::numpy_iexpr<(anonymous
namespace)::pythonic::types::numpy_iexpr<const (anonymous namespace)::pythonic::types::ndarray<double, (anonymous namespace)::pythonic::types::pshape<long, long, std::integral_constant<long, 2>>>
&>>, (anonymous namespace)::pythonic::types::contiguous_normalized_slice> &, (anonymous namespace)::pythonic::types::list<long>, 1, 0, false>' requested here
return broadcast_copy_helper<E, F, N, D, vector_form>(
I'm also getting runtime errors :( divides by zero for example.
Your first example is enough for me to create a reproducer, thanks. We'll discuss the divide by zero aspect afterward :-)
Ah great, let me know when I can dig further. I think the other issues are related to the fact that an integer scalar is not getting updated from 0, I think I see that in a lot of places, and that could also be the cause for the / 0 ...
After pulling the latest version I get this error:
INFO: gcc: /var/folders/j6/7t8sg1vj4q97zhh9z5cdmxbm4rz935/T/tmpba0fmb0u.cpp
In file included from /var/folders/j6/7t8sg1vj4q97zhh9z5cdmxbm4rz935/T/tmpba0fmb0u.cpp:1:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/core.hpp:46:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/slice.hpp:5:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/NoneType.hpp:6:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/builtins/bool_.hpp:6:
In file included from /Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/tuple.hpp:13:
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/types/ndarray.hpp:1058:5: error: reference to non-static member function must be called
data = utils::shared_ref<T>(other.begin(), other.end());
^~~~
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/include/types/list.hpp:330:12: note: possible target for call
dtype* data() { return _data->data();}
^
/Users/jlaroche/ua_conda_env-Darwin.x86_64/lib/python3.8/site-packages/pythran/pythonic/include/types/list.hpp:331:18: note: possible target for call
const dtype* data() const { return _data->data();}
Obsoleted by #2098
This is faster for scalars, but not for objects, so we need to generate std::move accordingly