23 Mar 2018 |
| Matthew Wardrop changed the history visibility to "world_readable" from "shared". | 13:05:32 |
| Matthew Wardrop invited danfrankj. | 20:43:26 |
| Matthew Wardrop set a profile picture. | 20:48:18 |
| danfrankj joined the room. | 21:04:16 |
24 Mar 2018 |
| danfrankj set a profile picture. | 00:18:17 |
17 Apr 2018 |
danfrankj | Matthew Wardrop: question, for mensor, do measure providers need to know about external measures / via_names, etc? I wonder if there's a way for them to operate under a simplified contract. | 18:29:53 |
Matthew Wardrop | In order to handle the intermediate representation from other compatible measure providers, I think they do. | 19:26:31 |
Matthew Wardrop | I've just pushed in my SQL implementation, btw. | 19:26:40 |
Matthew Wardrop | Did you see it? | 19:26:46 |
danfrankj | taking a look now... does it handle subject aggregations? | 20:39:21 |
Matthew Wardrop | Subject meaning dimensions? | 20:39:58 |
Matthew Wardrop | The only thing it doesn't currently handle is constraints. | 20:40:10 |
Matthew Wardrop | I'll get to that soon :) | 20:40:19 |
danfrankj | I mean, the 2-stage aggregations... e.g. you might want to compute "orders" and "orderers" from the same source | 20:41:25 |
Matthew Wardrop | With some limitations, yes. | 20:42:11 |
Matthew Wardrop | Check out the tests/test_pandas_provider.py test/example | 20:42:46 |
Matthew Wardrop | Everything done in pandas there should work equally well with the SQL backend | 20:43:34 |
Matthew Wardrop | Missing functionality from all backends: proper support for constraints and multi-key joins (though this is now partially supported by the backend). | 20:44:42 |
Matthew Wardrop | The first of these is a top priority for me, so will be added soon. | 20:45:10 |
Matthew Wardrop | The latter is useful, but challenging and less immediately useful. | 20:45:48 |
danfrankj | just looked at the tests ... I don't see any example of "identifier aggregations" | 20:47:59 |
danfrankj | e.g. orderers by date | 20:48:26 |
Matthew Wardrop | So, e.g., the number of orders placed on a given date? | 20:50:13 |
Matthew Wardrop | m.evaluate('order', measures=['count'], where=['ds="2018-01-01"']) | 20:50:42 |