Include information about additionalProperties in object tables #1798
Conversation
richvdh
force-pushed
the
rav/additional_properties
branch
from
63fbe8d
to
b33a6a9
Compare
| * | ||
| * Returns a dict containing: | ||
| * | ||
| * * `objects`: The array of object schema definitions. | ||
| * | ||
| * * `schema`: An updated copy of the `schema` input (eg, nested `allOf` |
There was a problem hiding this comment.
Since #1787, allOf entries are now resolved at the top level before we get here, rather than inside resolve-additional-types.
Just documenting the current situation.
…ct-table` Previously, we were stripping `additionalProperties` and `patternProperties` from all objects except top-level objects. Obviously, this was no good for objects where a nested property contains such properties. Fixing that (in `clean-object`) *ought* to be simple enough, except that it turned out we were relying on the fact that would give us an "empty" entry in the array of types-to-render-tables-for returned by `resolve-additional-types`. (Normally, we don't want an object that only has `additionalProperties` to have its own table, since we just embed it in the parent table.) So, we need to add more logic to `resolve-additional-types-inner` to suppress such tables. This commit doesn't change the rendered output at all (verified via `diff`): it's just preparation for what comes next.
richvdh
force-pushed
the
rav/additional_properties
branch
from
b33a6a9
to
b020b1d
Compare
There was a problem hiding this comment.
I almost worked on that today, until I realized we needed a way to show patternProperties too, which is why I worked on #1796 first 😉.
There is at least one TODO comment to remove at:
Great minds think alike 😇. It's not the first time the two of us have ended up working on the same thing. I think we should probably try to mention what we're planning to work on in |
|
Sorry for leaving this to sit for so long. These changes LGTM. |
Currently, if we have an object which has
additionalPropertiesin addition toproperties, that information gets lost. This PR seeks to address that.The individual commits should be separately reviewable.
Examples:
old:
new:
old:
new
old
new
Some of those descriptions could use improvement, but that is a separate job.
Preview: https://pr1798--matrix-spec-previews.netlify.app