ElasticSearch refresh parameter unused?

[ccancellieri]

Describe the bug
The pagination does not work properly after massive items upload over ElasticSearch as the number of items should still be calculated by ES (probably).
After a while the number is correct and the pagination starts to work.

I noticed that in the driver we have the support for the refresh parameter correctly exposed but it is never used by the stac-api as it is an abstraction layer while 'refresh' is something specific for ElasticSearch:

F.e.:

    async def delete_item(
        self, item_id: str, collection_id: str, refresh: bool = False
    ):
        """Delete a single item from the database.

        Args:
            item_id (str): The id of the Item to be deleted.
            collection_id (str): The id of the Collection that the Item belongs to.
            refresh (bool, optional): Whether to refresh the index after the deletion. Default is False.

        Raises:
            NotFoundError: If the Item does not exist in the database.
        """
        try:
            await self.client.delete(
                index=index_by_collection_id(collection_id),
                id=mk_item_id(item_id, collection_id),
                refresh=refresh,
            )
        except exceptions.NotFoundError:
            raise NotFoundError(
                f"Item {item_id} in collection {collection_id} not found"
            )

But from the stac-api refresh is never passed. Am I wrong?

Should we hardcode 'wait_for' or at least use the kwargs in some way?

[jonhealy1]

Hi @ccancellieri did you ever think of a good solution for this?

[jonhealy1]

In this pr, refresh can be passed via kwargs, or set globally for all transaction methods #370

[jonhealy1]

reopen if needed