compiler: Fix dtype of FFP and edit sympy_dtype #2513
Conversation
| dtypes.add(i.dtype) | ||
| except AttributeError: | ||
| pass | ||
| for i in expr.args: |
There was a problem hiding this comment.
This will not work for nested expression, why does the previous one not work?
There was a problem hiding this comment.
It passes all tests I believe (ofc without the ascent import issue).. so I'm not sure what kind of expression you are referring to? Will the nested expression not have a dtype attribute? Maybe I need singledispatch?
There was a problem hiding this comment.
Searching through expr.free_symbols for types like FieldFromPointer is problematic because it examines the dtype of the pointer itself rather than the dtype of the field.
There was a problem hiding this comment.
then you may need to use singledispatch to implement sympy_dtype (which, btw, I'd rename into something else as "dtype" isn't even a SymPy thing; like, extract_dtype or something like that)
An alternative would be to turn it into a recursive function (check expr; if not enough, check its arguments and for each argument call the function recursively), but it might not be enough for special objects like FieldFromPointer
| @@ -292,7 +303,7 @@ def test_extended_sympy_arithmetic(): | |||
| # noncommutative | |||
| o = Object(name='o', dtype=c_void_p) | |||
| bar = FieldFromPointer('bar', o) | |||
| assert ccode(-1 + bar) == '-1 + o->bar' | |||
| assert ccode(-1 + bar) == 'o->bar - 1' | |||
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #2513 +/- ##
==========================================
- Coverage 87.28% 78.50% -8.79%
==========================================
Files 238 238
Lines 45703 45712 +9
Branches 4057 4058 +1
==========================================
- Hits 39892 35885 -4007
- Misses 5126 9078 +3952
- Partials 685 749 +64 ☔ View full report in Codecov by Sentry. |
FabioLuporini
changed the title
No description provided.