Releases: apibara/dna
Releases · apibara/dna
Parquet Sink v0.3.0
Introduce sink status gRPC service.
Changed
- The status server is now a gRPC service. This service returns the sink
indexing status, the starting block, and the chain's current head block
from the upstream DNA service. - The status server now binds on a random port. This means it's easier to run
multiple sinks at the same time.
Mongo Sink v0.4.0
Introduce sink status gRPC service.
Changed
- The status server is now a gRPC service. This service returns the sink
indexing status, the starting block, and the chain's current head block
from the upstream DNA service. - The status server now binds on a random port. This means it's easier to run
multiple sinks at the same time.
Console Sink v0.3.0
Introduce sink status gRPC service.
Changed
- The status server is now a gRPC service. This service returns the sink
indexing status, the starting block, and the chain's current head block
from the upstream DNA service. - The status server now binds on a random port. This means it's easier to run
multiple sinks at the same time.
Apibara CLI v0.2.0
Introduce sink status gRPC service.
Changed
- The status server is now a gRPC service. This service returns the sink
indexing status, the starting block, and the chain's current head block
from the upstream DNA service. - The status server now binds on a random port. This means it's easier to run
multiple sinks at the same time.
Postgres Sink v0.3.0
Bring support for TLS connections.
Changed
- Breaking: use TLS by default. You can revert to the old insecure
connection by using the--no-tls=true
CLI flag, or setting
sinkOptions.noTls: true
in your script.
Added
- You can now connect to PostgreSQL securely using TLS connections. The TLS
connection can be customized by providing a self-signed certificate, or by
enabling/disabling certificate and hostname validation.
Starknet v1.1.2
Add a new Status method to the gRPC service.
Added
- Add a
Status
method to theStream
gRPC service. This method is used to
query the current service ingestion state. - Add
--address
CLI flag to change on which address the DNA service listens
for connections. Defaults to0.0.0.0:7171
for backward compatibility.
Mongo Sink v0.3.0
Introduce entity mode to index stateful entities.
Added
- Add entity mode to index stateful entities. When this mode is turned on
(by setting theentityMode
option totrue
), the indexer behaviour changes
to enable updating entities while indexing.
The return value of the transform is expected to be a list of{ entity, update }
objects, whereupdate
is a MongoDB update document or pipeline.
Please refer to the document to read more about entity mode.
Webhook Sink v0.2.0
This release improves the developer experience when running locally.
Added
- Add a
--persist-to-fs=my-dir
flag to persist the indexer's state to the
filesystem. This option creates a new directory (specified by the user) and
writes the indexer's current state to a file, with one file per indexer.
Developers shouldn't use this option for production, but only for
development since it lacks any locking mechanism to prevent multiple copies
of the same indexer running at the same time.
Changed
- The transform function now is invoked with a single block of data.
- In "raw mode", each value returned by the transform function is sent as an
individual request.
Postgres Sink v0.2.0
This release improves the developer experience when running locally.
Added
- Add a
--persist-to-fs=my-dir
flag to persist the indexer's state to the
filesystem. This option creates a new directory (specified by the user) and
writes the indexer's current state to a file, with one file per indexer.
Developers shouldn't use this option for production, but only for
development since it lacks any locking mechanism to prevent multiple copies
of the same indexer running at the same time.
Changed
- The transform function now is invoked with a single block of data.
Batching is a low-level mechanism used to control how often data is written to Postgres,
but it shouldn't affect the transform step.
Parquet Sink v0.2.0
This release improves the developer experience when running locally.
Added
- Add a
--persist-to-fs=my-dir
flag to persist the indexer's state to the
filesystem. This option creates a new directory (specified by the user) and
writes the indexer's current state to a file, with one file per indexer.
Developers shouldn't use this option for production, but only for
development since it lacks any locking mechanism to prevent multiple copies
of the same indexer running at the same time.
Changed
- The transform function now is invoked with a single block of data.