WIP: feature: TracerBundle#235
Conversation
ea06af4 to
4e4dd6e
Compare
| kwargs[name] = arg.data | ||
| except AttributeError: | ||
| pass | ||
| def _convert_NDSL_concepts_to_storage(args: tuple, kwargs: dict) -> tuple[tuple, dict]: |
There was a problem hiding this comment.
Good catch.
Alternative is to have a ABC class and/or implement a __to_gt4py_storage__ functions that would be tested, like DaCe or numpy does for it's interface systems
There was a problem hiding this comment.
Yeah, this is really just to make it work. I could see something like a GT4PyStorageConvertible base class or so that would expose a function convert the concept to something that GT4Py understands. I'll come back to that once we have a working prototype (including orchestration).
There was a problem hiding this comment.
Can also be a "refactor later" tasks, it's in the back enough to be "safe"
| be derived from a Storage if given as the data argument, otherwise the | ||
| storage attribute is disabled and will raise an exception. Will raise | ||
| a TypeError if this is given with a gt4py storage type as data | ||
| allow_mismatch_float_precision: ??? |
There was a problem hiding this comment.
By default allocating any floating point fields is tested against the system-wide Float. This attribute allows user to flag that their usage of np.float32/64 is needed (defensive programming)
There was a problem hiding this comment.
good to know, I'll add this in the final cleanup (once things are working)
3a12ffb to
40e1740
Compare
5b2acb5 to
25476da
Compare
Test isn't passing. This isn't really working yet ...
Note: I haven't checked validation ...
d4d86e3 to
de04821
Compare
Description
Initial attempt to write a DSL-like, orchestrateable version of a
TracerBundle.How has this been tested?
Please describe the tests that you ran to verify your changes. Please also note
any relevant details for your test configuration (e.g. compiler, OS). Include
enough information so someone can reproduce your tests.
Checklist