public / sequelize

* Add `isSoftDeleted` helper method to model instances

Since users can rename the `deletedAt` field when creating a model it
is helpful to have a `isSoftDeleted` so other users of the DB don't
have to hard code the custom `deletedAt` field name throughout their
codebase.

Addresses: https://github.com/sequelize/sequelize/issues/7408

* Update `isSoftDeleted` to check for dates in the future

If the soft-delete property (default is `deleted_at`) is set, but it's
set to a date in the future, then `isSoftDeleted` should return `false`.

* Update `isSoftDeleted` to return `true` if `deletedAt` is now

* Rename variable `isNotSet` to `isSet` and inverse logic

This helps with readability and understanding.

* [MSSQL] Format isolation level as tedious isolation level

* [ci skip] Added changelog entry

* Added unit test that stubs beginTransaction method

* Fixes jshint error

* Explicitly pass tedious lib instance

http://docs.sequelizejs.com/en/v3/docs/getting-started/ link is not active. Thanks

* Indent Markdown by 2 spaces

Consistent with most examples and our codebase

* Use ES6 (as supported by Node 4) in docs

Add Yarn installation scripts to getting started documentation. With many developers making use of Yarn as of late, providing quick copy/paste scripts will prevent developers from having to type the scripts manually.

* Two birds with one stone.  Adds support for ordering by JSON columns and removes vulnerability for injections on JSON attributes.

* Updates the changelog for the addition of json attributes.

* Fixes lint error with prefer arrow.

* Removes specific error because it may change depending on dialect and engine version.

* Removed duplicate tests and moved location of SQL injection test and find order test.

* Optimizes item.split performance by only performing the split once.

* Fixes test for sqlite dialect json.  JSON is handled differently in sqlite where the json value is extracted from a path.  If the path doesn't exist sqlite returns null instead of throwing an error like postgres.

* Removes support for $raw

* Removes interpretation of where array as a literal with replacements.

* Updates documentation to remove where object with literal replacements.

* Fixes some of the integration tests.

* Fixes integration tests and encapsulate subquery join where in a plain object.

* Adds unit tests for nested arrays in where object

* This should cover the rest of the diff.

* [ci skip] add BC break note to changelog

* Fixes changelog typo and removed the ternary because I edit in an abstract fashion.

* Fix upsert behavior to update all _changed_ fields by default

* Update changelog

* slice the options argument if the logging is set to console.log

* added test case to check if the options object is available for custom logging function, this test case has been added for the pull request 'slice the options argument if the logging is set to console.log (#7555)'

Previous one was giving 404.

* docs: first esdoc steps

* remove: deprecated articles

* fix: a bit of indentation

* add: router with redirects for surge

* fix: small css fixes (hide docs links). fix docs command

* add domain to travis.yml

* chore(package): remove unused packages, add esdoc

* fix: lint

* fix: links in /docs + other small changes

* shields

* remove eslint comments. Add back UnknownConstraintError

* Manual eslint fixes in test

* Manual eslint fixes in lib

* 🔥 remove jshint

* eslint --fix

* 'clarifying the return value of findOrCreate'

The documentation of findOrCreate previously had no precise indication of what to expect findOrCreate to return. I added that it returns an array with two parts in the comments that had already existed in the examples.

* minor typo fixes

* Made some wording changes for tense

* Fix `showIndex()` query regexp

* [ci skip] changelog for #7492

* Added add, remove & show constraints feat

* Added showConstraintsQuery for postgres & mysql

* [ci skip] Added docs for constraints usage

* Added removeIndex support for mysql

* Added altering constraints to sqlite dialect

* Refactor onDelete, onUpdate to work in sqlite dialect

* Throw non existent constraints in removeConstraint query as UnknownConstraintError, improve test coverage

* Fix failing tests

* Removed dialect naming check in favor of dialect.supports feature check

* Improve code coverage, eslint fixes

* Review fixes

* Review fixes

* Renamed model, key to table, field for foreign key contraint definition

* run hooks during validation by default

* Revert "run hooks during validation by default"

This reverts commit da66b72bce4bc274f10745ec98cad86ccb1def20.

* hooks by default

* write validation sequence more cleanly and migrate instance validator integration tests to unit tests

* remove only

* ignore w30

* change log

* fix jshint error and use const

* jshint fix

* only validationFailed  hook when validation fails

* fix changelog

* fix changelog

* add integration tests back

* fix indentation

* and fix spacing

* final code review changes

https://github.com/sequelize/sequelize/issues/7404

* Added feature to alter tables on sync. Fixes #537

* Added an entry to future in changelog

* Added integration test for syncing with alter

* Fixed tests and model with sqlite and psql

* Updated docs and updated some code style

* Updated test.sync.js to use let instead of var

* Changed all vars to lets and replaced anonymous functions with arrow functions

* Added back some anonymous functions to use scope and fixed more coding style

* Updated code style again - removed binds and nested then's

* Updated changelog; renamed and improved tests; Added option to comment

* Changed debug vars to consts

* Switched changes to const

* Added some missing semi-colons

* issue # 7318

fixes issue # 7318 - https://github.com/sequelize/sequelize/issues/7318

* test and change log

* formatting

* changed message

* Update changelog.md

* Update removeAttribute.test.js

In generating the bulk insert query, the serials are stored in an
array. This array can potentially be up to size N, which is the number
of rows to be created.

When generating the tuples at a later stage, we walk through the
records one-by-one - O(N) iterations. However, in a single iteration,
we were performing a lookup in the serials array (indexOf costs O(N)),
leading to a O(N^2) runtime and poor performance for large number of
records inserted.

This patch converts the serials array into an object map instead, so
lookups cost O(1) time.