0.6.1

Summary

Workaround for cassandra storageType field-resolvers. See related issue / PR

Issues

• #186

PRs

• #188

0.6.0

0.6.0

Changelog

storage types

Added support for the following storage types

• cassandra
• mongodb
• amazonS3
• presto / trino
• neo4j

Domain specific language

This Version implements a change in the zendro domain specific language to define data models, specifically their associations between each other.

• With this version there are 4 types of possible association types:

  • one_to_one
  • many_to_one
  • one_to_many
  • many_to_many

• Additionally two additional fields were added.

  • implementation. Can be one of foreignkeys, generic, sql_cross_table. Describes the implementation on the model level
  • reverseAssociation. Name of the reverse association. Can be omitted for codgen of the server. Needed in the SPA.

• The field reverseAssocitionType for many_to_many via foreign key arrays was removed. If an association is of type many_to_many and the implementation is foreignkeys it will be done via foreignkey arrays.

• The fields keyIn in and keysIn were unified to field keysIn

Connection Helper

For Connection type resolvers implments direct access to the record without going through edges { node }

Merged Pull requests

The following Pull requests are merged

#172 #181 #182 #183 #184 #185 #187

Version 0.5.0

Version 0.5.0

PRs

gql-codegen#157
gql-codegen#158
gql-server#49
gql-spa#18
gql-spa-codgen#64
graphiql-auth#7
starterPack#11

Changelog

graphql-server-model-codegen

• use zendro-env to manage the integration-tests environment
• added -m flag to codegen for optional generation of the migrations

refactor readMany API to not use counts and have mandatory pagination

• pagination is mandatory!.

  • limit-offset: paginationInput and limit are required by the schema.
  • cbp: paginationCursorInput required by schema. Either first or last has to be given. This is checked in the resolver. Error if invalid.

• resolver checks record limit, model-layer does not care

  • checkAndAdjustRecordLimit resolver function removed, pagination arguments used instead. If exceeded error is thrown. This is now also the case in DDMs

• views/includes/record-limit-helpers.ejs removed as deprecated

• in case of to_one end of an association pagination with LIMIT 1 (either limit or first) is given to the root-resolver call.

• in case of one_to_one end of assoc without the foreignKey pagination with LIMIT 2 is used in the root resolver, to check if a record is associated to more than 1 record. Appropriate benignError is added.

• readAll and readAllCursor refactored:

  • neither does need to count.
  • neither will use a limit, if no limit is given. (e.g. when requiring a model manually inside node)
  • readAllCursor validity check of pagination args moved to resolver.
  • To avoid calculating the OFFSET needed for backward-pagination, instead the resulting records will be reversed post-fetching to get the desired order.
  • To calculate the pageInfo (hasNextPage | hasPreviousPage) the query limit (if needed) will be increased by 1 (Extra check needed for the special case someone uses pagination:{first: -1}), to get information about possible succeeding records.
  • Additionally a second query with reversed Operators (<,<=,>,>=) with no ORDER and LIMIT 1 will be run to get possible preceding records.
  • Functionalities outsourced to helper functions.

• changed error message for invalid response from remote zendro-server.

• Tests

  • Integration Test "21. Limit check" refactored
  • added Tests for DDM and Zendro-webservice
  • added integration test for generic readAllCursor

graphql skeleton

• clean up server.js

• add Helpers for readAll & readAllCursor:

// SQL
searchConditionsToSequelize(search)
orderConditionsToSequelize(order, idAttribute)
orderConditionsToSequelizeBefore(order, idAttribute)
cursorPaginationArgumentsToSequelize(pagination, sequelizeOptions, idAttribute)
buildLimitOffsetSequelizeOptions(search, order, pagination, idAttribute)
buildCursorBasedSequelizeOption(search, order, pagination, idAttribute)
buildOppositeSearchSequelize(search, order, pagination, idAttribute)
buildPageInfo(edges, oppRecords, pagination)
buildEdgeObject(records)
reverseOrderConditions(order)

// Generic
cursorPaginationArgumentsToGeneric(search, pagination, orderOptions, idAttribute)
buildCursorBasedGenericOptions(search, order, pagination, idAttribute)
buildOppositeSearchGeneric(search, order, pagination, idAttribute)

• rename checkCountAndReduceRecordLimitHelper to checkCountAndReduceRecordsLimit

single-page-app-codegen

• added non-blocking counter for table view and association lists
• refactor non-associated lists for mandatory pagination
• refactor plotly request module.
• refactor to-one display of the currently associated records
• refactor: simplify request module

single-page-app skeleton

• add global environment variable config file
• add logRequest helper for displaying of the request
• add checkResponse helper to check request response for Errors
• refactor getSearchArgument to be generic helper function

documentation

• added documentation on cursor-based-pagination argument
• added mandatory pagination argument to all places in the docu where needed

0.4.1

graphql-server-model-codegen

• Fix unit tests to use the internal sequelize model instance introduced in release 0.3.0.

• Refactor search input value from a composite object to two separate fields withing the search arguments: value and valueType.

  input search<Model>Input {
    field: AccessionField
    value: String
    valueType: InputType
    operator: Operator
    search: [search<Model>Input]
  }

• Refactor valueType to a new enum InputType that accepts all GraphQL primitives supported at Zendro, and a new Array type used only by the search resolver.

  enum InputType{
    Array
    String
    Int
    Float
    Boolean
    Date
    Time
    DateTime
  }

graphql-server

• Refactor the search input value spec in static files to match the changes in `graphql-server-model-codegen.

single-page-application-codegen

• Refactor model requests to use the new value spec in graphql-server and graphql-server-model-codegen.
• Fix integration tests to use the new configuration changes introduced in release 0.3.0

0.3.0

graphql-server:

• Added support for arbitrary local database connections in the data-model definitions (only sql support at the moment).

  • A new optional database property in the data-model definition specifies the connection it should use.
  • If this property is not defined, the mandatory storageType property value will be used to get the default configuration.
  • Each connection creates a shared sequelize instance. Separate database migrations and seeding will be performed on each connection.
  • The seeders folder must now have a sub-folder structure that matches the database name that needs to be seeded (e.g. seeders/default-sql/<seeders-file.js>).

• The default sequelize configuration file config.json has been removed in this release. A new data_models_storage_config.json file is included.

  • This file should include all connections required by the graphql-server.
  • The default connection for any data-model is specified as default-<storageType>. This <storageType> suffix should match the storageType property in the data-model (e.g. default-sql).
  • A new property storageType is also required in each connection (e.g. { storageType: "sql" }).
  • See the included data_models_storage_config.json for an example of the default connection.

graphql-server-model-codegen

• Code-generator changes to support the new arbitrary database configurations.

  • Migrations create a sub-folder structure matching the keys in data_models_storage_config.json.
  • Adapted generated models to use the new shared connection pool.
  • Each generated model now uses its internal sequelize instance to perform operations.

• Minor changes to the integration tests command-line interface.

  • Added -b <branch> flag to dynamically checkout and fetch remote branches in the test graphql-server instances. This flag can be used during a default run, or in combination with -k, -s, and -T flags.
  • Added -C flag to only remove containers and generated code.
  • Added a -s flag to only setup or reset the testing environment server instances.
  • These changes allow running multiple branch tests without having to rebuild the Dockerfile.graphql_server image, and provide a more fine-grained control over the testing step process (e.g. for debugging).

• Extensive refactoring of the integration tests API:

  • Adapted tests to work with the new configuration changes.
  • Refactored functions to be more modular and provide a more composable API.
  • Detached docker image builds from graphql-server installation. This allows manipulation of the server instances without having to rebuild images.
  • Deferred jq binary compilation to the server image build, which is used by node-jq.
  • Standardized debug logs and made optimizations where possible.