v1.2.0 - 2024-10-25
-
Change mutation pre/post checks to be optional
-
Added
mutationPrefix
to configuration allowing mutation names to be customised.
v1.1.2 - 2024-10-07
- Enabled point mutations by default for newly generated configuration.
- Make introspection FK column order deterministic, preventing incorrect composite key column mapping
v1.1.1 - 2024-08-22
- Support filtering by a nested field collection. #573
- Support setting ssl client certificate information and ssl root certificate independently. #578
v1.1.0 - 2024-08-16
- Support setting ssl client certificate information via environment variables. #574
- Make array element types nullable in the schema. #571
- Upgrade ndc-sdk-rs to v0.3.0, including the following changes:
- Listen on all IPv4 and IPv6 changes by default ndc-sdk-rs#22
- Replace the health check with a readiness check ndc-sdk-rs#27
- Allow unsecured requests to the
/health
endpoint ndc-sdk-rs#27
- Make fields of composite types nullable in the schema. #565
- Allow Native Operations that end with a semicolon when it's easy to remove them. #566
- Fix nested field relationships. #564
- Support network_supeq and network_subeq by default. #541
- Generate the comparison operator
_neq
. #540
- Introduce configuration version "v5". #522
- Support ndc-sdk-rs v0.2.1, including changes to error messages. #520
- Warn when starting the connector with an older configuration version. #537
- Add configuration version metrics. (#504)
- Introduce mutationsVersion 'v2' which auto-generates update, delete, and insert point mutations. (#513)
- Implement native query builder in the CLI plugin. (#511)
v0.7.1 - 2024-06-20
-
A default value for
CONNECTION_URI
for newly initialized connector configurations. (#503) -
Support ndc-spec v0.1.4. (#499)
-
Support ndc-spec v0.1.3 and filter/order by nested fields. (#486)
- Aggregate functions (except COUNT) are now all marked as having nullable return types, as they will return null for empty row sets. (#498)
- Rows and aggregates parts of the query should operate on the same query parameters (where, order by, limit and offset). (#471)
v0.7.0 - 2024-05-22
- Support for qualifying scalar types by their schema. This updates the
metadata configuration format version number from
"3"
to"4"
. (#471)
- When acquiring a connection, ping the db only if it has been idle. (#450)
- Return at most a single instance of a root table row when filtering, empty rows of a column relationship should not necessarily filter a row. (#463)
v0.6.0 - 2024-04-16
- Support nested field selection. (#404) (#409)
- Expose the type representation of base types and domain types. (#398)
- Expose the type representation of enums via the ndc schema. (#397)
- Support ndc-spec v0.1.2 and change the type representation of types accordingly. (#408)
int8
andnumeric
columns will now emit a string json representation by default. (#416)- Enforce read-onlyness without transactions. (#421)
- Fix schema conflict result_type for native query mutations. (#405)
- The CLI plugin no longer requires the Visual C++ Redistributable on Windows. (#417)
- Include additional schemas for types by default. (#420)
- Remove the 'Caused by' section in the error message of the cli. (#424)
v0.5.2 - 2024-03-29
- Make operators of domain types available via implicit casts. (#392)
- Support introspection composite types. (#391)
- Support enum types in the configuration. (#387)
- Make aggregation functions available through implicit casts in the configuration. (#381)
- Support for introspecting domain types. (#380)
- Support ndc-spec 0.1.1. (#390)
v0.5.1 - 2024-03-21
- Support for reading Native Query SQL from files. (#372)
- Do not print information about when the cli decides not to write to a file. (#365)
- Refer to the right column name in the schema endpoint. (#366)
- The CLI now supports older versions of glibc, so is more likely to work on older operating systems (e.g. Ubuntu 20.04 Focal). (#373)
v0.5.0 - 2024-03-14
- The CLI plugin now integrates with the DDN CLI's
watch
functionality. (#360)
-
The default connection pool settings will now be written to the configuration file upon initialization (#351)
-
The ndc-postgres-cli
initialize
command will now generate the jsonschema of the configuration format as well. (#361) -
The ndc-postgres-cli
update
command will check that the configuration file has not changed in the middle of the introspection process, will retry if it did, and will not write the introspection result to file if the configuration file is already up to date. (#362) -
A few fields in the configuration format has been changed:
configureOptions
was renamed tointrospectionOptions
connectionUri
,poolSettings
andisolationLevel
are now nested underconnectionSettings
mutationsVersion
was moved fromconfigureOptions
to the top-level- A new field was added:
$schema
references the jsonschema of the configuration format
(#361)
v0.4.1 - 2024-03-06
- The Docker image's default command is now
serve
so it can be run without arguments, in line with the connector specification. (#345) - The connector manifest has been fixed to refer to the CLI plugin by name. (#344)
v0.4.0 - 2024-03-04
- There is now a CLI plugin for the Hasura v3 CLI, which allows the CLI to automatically introspect the database on demand. (#307 and #312)
- Support for explaining mutations via the
/mutation/explain
endpoint. (#283) - Support filtering using
in
by columns and variables. (#283)
- We have upgraded to ndc-spec v0.1.0, which demands a few changes:
- Version 1 and 2 of the configuration and NDC metadata have been deprecated and removed.
- Configuration is now a file,
configuration.json
, in the specified configuration directory. It is now version 3, but little has changed from version 2. This brings ndc-postgres in line with the connector specification. (#305) - When configuration is initialized, it is expected that the connection URI will now be specified using an environment variable, instead of written directly. This can be overridden. (#325)
- The default port was changed from 8100 to 8080. This is configurable with
the
HASURA_CONNECTOR_PORT
environment variable. - Types and procedures are only supported in unqualified schemas, specified in
unqualifiedSchemasForTypesAndProcedures
. (#271) - The configuration server has been removed in favor of a cli interface. (#307 and #312)
- Fix queries including an IN operator on an empty list. (#309)
v0.3.0 - 2024-01-31
- Basic support for auto-generated insert procedures for tables. (#261)
- Support for composite types in inputs and outputs. Right now, no operators are provided for composite types, and there is no automatic introspection. (#240)
- Support for composite types and arrays in variables. (#249) (#263)
- When explaining a query without variables (such as when joining across a
remote relationship), explaining now still provides the generated SQL, and
omits the
EXPLAIN
result from PostgreSQL. (#241)
- Native query mutations are now marked as procedures instead of collections (#234)
- Introspection now avoids
DISTINCT ON
to be more compatible with PostgreSQL variants (#260) - Invalid request, constraint not met, and unprocessable content errors are emitted at the relevant scenarios (#239)
v0.2.0 - 2023-12-21
- Support for introspecting prefix-functions as comparison operators (#223)
- Support prefix functions as comparison operators (#220)
- Support queries without fields specified. (#209)
- Query transactions now run in read-only mode, and allow transaction isolation level configuration. (#209, #212)
- Support variables as arguments to native queries (#211)
- Introduce version 2 of connector deployment configuration. (#208)
- Support array types (#191, ...)
- Support Native Query Mutations (#189, #198, #222)
v0.1.0 - 2023-11-29
Initial release.