Oscar Benjamin

Would `flint_set_abort` handle this or is that not applicable to a memory error like this? I had been considering whether python-flint should use cysignals. There is some discussion on the...

> I've set it to raise an exception on any division function with a non-prime modulus, even if it be within the domain. The same goes for sqrt and gcd....

> I think that's also reasonable, currently I'm just checking the first time, storing the result and just returning that instead for later calls. Either approach is reasonable I think....

> > This is why we need a context for nmod so that we can store whether or not the modulus is prime and only do the prime check once....

The test failure is ``` Traceback (most recent call last): File "/opt/hostedtoolcache/Python/3.12.4/x64/lib/python3.12/site-packages/flint/test/__main__.py", line 36, in run_tests test() File "/opt/hostedtoolcache/Python/3.12.4/x64/lib/python3.12/site-packages/flint/test/test_all.py", line 2888, in test_mpolys assert P(ctx=ctx).is_zero() is True ^^^^^^^^^^^^^^^^^^ AttributeError: 'flint.types.fmpz_mod_mpoly.fmpz_mod_mpoly'...

> Seems like CI was doing something and merging the PR before testing, definitely through me off for a minute. That is usually how CI works at least for Travis...

> Would this be replacing the global `FlintContext`? Instead moving it towards something instance specific? These are details that need to be worked out. For `arf` for example I would...

There should be some handling equality checking against scalars: ```python In [26]: flint.fmpz_poly([1]) == 1 Out[26]: True In [27]: ctx = flint.fmpz_mpoly_ctx(2, flint.Ordering.lex, ['x', 'y']) In [28]: ctx.from_dict({(0,0):1}) Out[28]: 1...

> This took a while to catch but I believe it's been fixed. The `ctypedef unsigned long ulong` in `flintlib/flint.pxd` causes Cython to inline the definition of `ulong` in place...

> I think its preferable to remove it or use the true definition, though I'm unsure how that would be done I think the only way to use the true...