When we migrate the current contents of the search index, we're going to be effectively overwriting each entry in place with the same content in a slightly updated schema. Testing on dev/staging revealed that this was indeed a breaking change (sorry!)
The important change is that entrypoints no longer have a steps field, but we didn't lose any information (since none of the existing entrypoints had more than one step), it's just at the top level of the entrypoint schema. gardens didn't lose any fields.
entrypoint.steps[0].function_name is still at entrypoint.short_name
entrypoint.steps[0].function_text is now at entrypoint.function_text
entrypoint.steps[0].description is still at entrypoint.description
For reference, the old schema for search index entries was the one used for the now-deprecated POST /garden-search-record route (docs) and the new schema is the one returned by the GET /gardens/{doi} route (docs)
There were a couple of additive changes to the schemas too (id and owner_identity_id fields for gardens and entrypoints), but those aren't likely to be the problematic ones
When we migrate the current contents of the search index, we're going to be effectively overwriting each entry in place with the same content in a slightly updated schema. Testing on dev/staging revealed that this was indeed a breaking change (sorry!)
The important change is that entrypoints no longer have a
stepsfield, but we didn't lose any information (since none of the existing entrypoints had more than one step), it's just at the top level of the entrypoint schema. gardens didn't lose any fields.entrypoint.steps[0].function_nameis still atentrypoint.short_nameentrypoint.steps[0].function_textis now atentrypoint.function_textentrypoint.steps[0].descriptionis still atentrypoint.descriptionFor reference, the old schema for search index entries was the one used for the now-deprecated
POST /garden-search-recordroute (docs) and the new schema is the one returned by theGET /gardens/{doi}route (docs)There were a couple of additive changes to the schemas too (
idandowner_identity_idfields for gardens and entrypoints), but those aren't likely to be the problematic ones