Update dev docs on how to use new object model #1054
Conversation
achantavy
requested review from
ryan-lane,
ramonpetgrave64,
csanders-git,
marco-lancini,
kedarghule and
mpurusottamc
| r.lastupdated = $UpdateTag | ||
| ``` | ||
|
|
||
| #### Connecting different node types with the `_attach` pattern |
There was a problem hiding this comment.
I feel like their may be cases where we want to attach after loading. But in the cases I can think of, it's still best to transform the data and use a NodeSchema that properly defines those extra relationships.
There was a problem hiding this comment.
Good point, and yes that's the approach I would encourage.
It's also possible to first generate a query using selected_relationships and then load to that with additional data, but I don't have an example ready for that and that sounds complicated and I'm not ready to encourage that pattern in the doc yet.
| As another example, `region: PropertyRef = PropertyRef('Region', set_in_kwargs=True)` tells the querybuilder to set a field called `region` on the `EMRCluster` node using a keyword argument called `Region` supplied to `cartography.client.core.tx.load_graph_data()`. `set_in_kwargs=True` is useful in cases where we want every object loaded by a single call to `load_graph_data()` to have the same value for a given attribute. | ||
|
|
||
|
|
||
| #### Defining relationships |
There was a problem hiding this comment.
I think it's worth adding an example here on the other_relationships, even if it's only pointing to the test cases we have.
* Update dev docs on how to use new object model * linter * example of other_relationships
Update the how-to-write-an-intel-module docs using the new cartography object model introduced in #1038.