Why is ThrottlingException not caught and retried by Connection._make_api_call:?

[ilanjb]

I'm using TransactWire to make batch updates
I'm getting this traceback

  File "/python3.7/site-packages/pynamodb/transactions.py", line 31, in __exit__
    self._commit()
  File "/python3.7/site-packages/pynamodb/transactions.py", line 132, in _commit
    return_item_collection_metrics=self._return_item_collection_metrics,
  File "/python3.7/site-packages/pynamodb/connection/base.py", line 1121, in transact_write_items
    raise TransactWriteError("Failed to write transaction items", e)
pynamodb.exceptions.TransactWriteError: Failed to write transaction items

I see that this is intentionally not retried
pynamodb.connection.base.Connection._make_api_call:
except (ValueError, botocore.exceptions.HTTPClientError, botocore.exceptions.ConnectionError) as e:
It seems that ThrottlingException should be retried. It is retried by default in boto3

[ikonst]

There's been work underway by @garrettheel to switch us to call boto under the hood, which would lead us to leverage boto's default behavior.

[ilanjb]

I'll follow there. thanks!