spacedrive/core/crates/sync
Brendan Allan 735e80ad4d
[ENG-1776] Delete cloud ops after they've been ingested (#2512)
* delete cloud ops after they've been ingested

* give wait_tx to p2p sync ingest

* remove Ingested event

* add sync docs for setting relation fields

* Update core/crates/sync/README.md

Co-authored-by: Oscar Beaumont <oscar@otbeaumont.me>

---------

Co-authored-by: Oscar Beaumont <oscar@otbeaumont.me>
2024-05-31 06:54:22 +00:00
..
src [ENG-1776] Delete cloud ops after they've been ingested (#2512) 2024-05-31 06:54:22 +00:00
tests [ENG-1776] Delete cloud ops after they've been ingested (#2512) 2024-05-31 06:54:22 +00:00
Cargo.toml Sync settings page (#2460) 2024-05-07 07:39:22 +00:00
README.md [ENG-1776] Delete cloud ops after they've been ingested (#2512) 2024-05-31 06:54:22 +00:00

sd-core-sync

Spacedrive's sync system. Consumes types and helpers from sd-sync.

Creating Records

Prepare a sync id by creating or obtaining its value, and then wrapping it in the model's SyncId struct, available at prisma_sync::{model}::SyncId.

Next, prepare the sync operations using some varaints of the sync_entry macros. sync_entry and option_sync_entry take the value first, and then the path to the field's prisma module. sync_db_entry and option_sync_db_entry take the same inputs, but additionally produce a prisma operation in a tuple with the sync operation, intended to be put into a Vec and unzipped.

Finally, use sync.shared/relation_create depending on if you're creating a standalone record or a relation between two records, and then write it to the database with write_ops.

let (sync_params, db_params): (Vec<_>, Vec<_>) = [
  sync_db_entry!(self.name, tag::name),
  sync_db_entry!(self.color, tag::color),
  sync_db_entry!(false, tag::is_hidden),
  sync_db_entry!(date_created, tag::date_created),
]
.into_iter()
.unzip();

sync.write_ops(
  db,
  (
    sync.shared_create(
      prisma_sync::tag::SyncId { pub_id },
      sync_params,
    ),
    db.tag().create(pub_id, db_params),
  ),
)

Updating Records

This follows a similar process to creation, but with sync.shared/relation_create.

let (sync_params, db_params): (Vec<_>, Vec<_>) = [
  sync_db_entry!(name, tag::name),
  sync_db_entry!(color, tag::color),
]
.into_iter()
.unzip();

sync.write_ops(
  db,
  (
    sync.shared_update(prisma_sync::tag::SyncId { pub_id }, k, v),
    db.tag().update(tag::id::equals(id), db_params);
  )
)

Deleting Records

This only requires a sync ID.

sync.write_op(
  db,
  sync.shared_delete(prisma_sync::tag::SyncId { pub_id }),
  db.tag().delete(tag::id::equals(id));
)

Relation Records

Relations require sync IDs for both the item and the group being related together. Apart from that they're basically the same as shared operations.

let (sync_params, db_params): (Vec<_>, Vec<_>) = [
  sync_db_entry!(date_created, tag_on_object::date_created)
]
.into_iter()
.unzip();

sync.write_ops(
  db,
  (
    sync.relation_create(
      prisma_sync::tag_on_object::SyncId {
        tag: prisma_sync::tag::SyncId { pub_id: tag_pub_id },
        object: prisma_sync::object::SyncId { pub_id: object_pub_id },
      },
      sync_params
    ),
    db.tag_on_object().create(
        object::id::equals(object_id),
        tag::id::equals(tag_id),
        db_params
    )
  )
)

Setting Relation Fields

Setting relation fields requires providing the Sync ID of the relation. Setting the relation field's scalar fields instead will not properly sync then relation, usually because the scalar fields are local and disconnected from the Sync ID.

let (sync_params, db_params): (Vec<_>, Vec<_>) = [
	sync_db_entry!(
		prisma_sync::object::SyncId { pub_id: object_pub_id },
		file_path::object
	)
].into_iter().unzip();

sync.write_ops(
	db,
	(
		sync.shared_update(
			prisma_sync::file_path::SyncId {
				pub_id: file_path_pub_id
			},
			sync_params
		),
		db.file_path().update(
			file_path::id::equals(file_path_id),
			db_params
		)
	)
)