Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Breaking changes regarding setting a default datetime in v0.15.1 #2715

Closed
maximedupre opened this issue Jul 13, 2018 · 9 comments
Closed

Breaking changes regarding setting a default datetime in v0.15.1 #2715

maximedupre opened this issue Jul 13, 2018 · 9 comments

Comments

@maximedupre
Copy link

maximedupre commented Jul 13, 2018

Environment

Knex version: v0.15.0
Database + version: mysql v5.7.18
OS: macOS 10.13.5

Bug

This worked in v0.15.0

exports.up = knex => {
    let createTable = (table) => {
        table.increments('id');
        table.integer('user_id').notNullable().unsigned();
        table.dateTime('creation_date').notNullable().defaultTo(knex.fn.now());
        table.dateTime('modification_date');
        table.string('share_token', 200);
        table.boolean('deleted').notNullable().defaultTo(false)
    }

    return knex.schema.createTable('projects', createTable)
};

But in v0.15.1

 create table `projects` (`id` int unsigned not null auto_increment primary key, `user_id` int unsigned not null, `creation_date` datetime(6) notnull default CURRENT_TIMESTAMP, `modification_date` datetime(6), `share_token` varchar(200), `deleted` boolean not null default '0') - Invalid default value for 'creation_date'
Error: Invalid default value for 'creation_date'
@tgriesser
Copy link
Member

This is due to #2542 which wasn't running in 0.15.0 and was fixed with a patch for 0.15.1, but looking closer this might not be what we want here.

@ricardograca I didn't look closely at the implementation of #2542, but since it's a configurable parameter and potentially breaking for things like this, I feel like it might be simpler to just make it an optional param:

.datetime('column', 6)

similar to other columns with precision such as string, float, decimal, etc.

@thetutlage
Copy link
Contributor

This also happens with the timestamp field.

Config

{
  client: 'mysql',
  version: '5.7',
  connection: {
    ...
  }
}

With the following code

table.timestamp('some_date').defaultsTo(knex.fn.now())

Returns the following error

Invalid default value for 'some_date'

@elhigu
Copy link
Member

elhigu commented Jul 14, 2018

the original issue about the reasons for the change is here #2524 basically the default behavior was pretty bad and inconsistent compared to other dialects.

@thetutlage
Copy link
Contributor

@elhigu Thanks for replying quickly. I am trying to get my head around what should be action item here.

So I am correct using knex.fn.now() is not the way to go with MYSQL>5.6?

@tgriesser
Copy link
Member

tgriesser commented Jul 14, 2018

@thetutlage @maximedupre if you specify the precision in the default to match the timestamp that'll be a quick fix. knex.raw("CURRENT_TIMESTAMP(6)") or knex.raw("NOW(6)") but ultimately this should be fixed - it's just a little tricky because the additional arguments in the builder are already used as a boolean to specify whether timestamptz should be used.

Working on a fix that accounts for both without breaking anything.

@ricardograca
Copy link
Member

ricardograca commented Jul 14, 2018

Right, the idea was to make MySQL behave like other dialects, that's why the precision wasn't optional. So in that same vein the knex.fn.now() method should probably be adapted as well to match the precision specification in case it's there.

@elhigu
Copy link
Member

elhigu commented Jul 14, 2018

@ricardograca was it right that currenty also setting version attr in config to 5.5 restores the old behaviour? Im on mobile right now so I dont want to browse too much, but IIRC plan was to add version attribute for now and keep old functionality as a default and change default behaviour to better precision to be default in 0.16?

@thetutlage
Copy link
Contributor

thetutlage commented Jul 14, 2018

Yes, setting 5.5 works as per the old behaviour. But it feels weird to use MySQL>5.6 and set version in the config as 5.5.

Again, I agree with the change, but just trying to understand is knex.fn.now() is way to go with the precision timestamp or I should simply stay away from using knex.fn.now()

@ricardograca

Yeah I agree that knex.fn.now() should be updated. However, I am not sure whether it's currently left out?

@thetutlage
Copy link
Contributor

I was looking into the source code and found that knex.fn.now() actually uses CURRENT_TIMESTAMP. https://github.com/tgriesser/knex/blob/232fe9f1517dba927f6a3a1fb1b8842d7c0a4007/src/functionhelper.js#L8

However, MYSQL also supports CURRENT_TIMESTAMP as a function, which takes the fraction. https://dev.mysql.com/doc/refman/8.0/en/date-and-time-functions.html#function_current-timestamp

Doing you think, we can add support for same?

FunctionHelper.prototype.now = function(precision) {
  return this.client.raw(`CURRENT_TIMESTAMP(${precision})`);
};

elhigu pushed a commit that referenced this issue Sep 26, 2018
* Kill queries after timeout for PostgreSQL

* Fix cancellation connection acquiring and test.

* Fix releasing connection in case query cancellation fails

* Add support for native enums on Postgres (#2632)

Reference https://www.postgresql.org/docs/current/static/sql-createtype.html

Closes #394

Signed-off-by: Will Soto <will.soto9@gmail.com>

* Removal of 'skim' (#2520)

* Allow overwriting log functions (#2625)

* Example build of custom log functions
* Handle logger object better for transactions
* Adjust test to ignore sqlite warning message

* Fixed onIn with empty values array (#2513)

* Drop support for strong-oracle (#2487)

* Remove babel-plugin-lodash (#2634)

While in theory, this may reduce the bundle size,
in practice it adds a ton of overhead during startup
due to the number of additional requires. Bundle
size also shouldn't matter for server side modules.

* Add json support to the query builder for mysql (#2635)

* Add json support to the query builder for mysql

refs #1036

Based on #1902

* Clarify supported version

* fix wrapIdentifier not being called in postgres alter column (#2612)

* fix wrapIdentifier not being called in postgres alter column

* add test for wrapIdentifier call in postgres alter column

* add comment regarding issue

* add issue & PR #'s in comment

* Remove readable-stream and safe-buffer (#2640)

* Add json support to the query builder for mysql

refs #1036

Based on #1902

* Clarify supported version

* Use native Buffer and Stream implementations

* fixes 2630 (#2642)

* Timeout errors shouldn't silently ignore the passed errors, but rather reject with original error. Fixes #2582 (#2626)

* chore: add Node.js 10 (#2594)

* chore: add Node.js 10

* chore: trigger new build

* chore: update lockfile

* chore: trigger new build

* fix: use npm i instead of npm ci

* Fix mssql driver crashing (#2637)

* Run SQL Server tests locally running SQL server in docker

* WIP mssql test stuff

* Patched MSSQL driver to not crash knex anymore

* Removed semicolon from rollback stmt for oracle (#2564)

* Remove WebSQL dialect (#2647)

* Add json support to the query builder for mysql

refs #1036

Based on #1902

* Clarify supported version

* Use native Buffer and Stream implementations

* Remove WebSQL dialect

* add homepage field to package.json (#2650)

* Make the stream catch errors in the query (#2638)

* Make the stream catch errors in the query

* Fix another case in which stream doesnt emits error

* Linter stuff

* Remove setTimeout in tests

* Make a test not to check the MySQL error code

* Fix stream error catching for MariaDB and PostgreSQL

* Fix stream error catching in Oracle

* Throw the error after emitting it to the stream

* Throw the error without instantiating a new Error

* Various fixes to mssql dialect (#2653)

* Fixed float type of mssql to be float

* Many tests where postgres test was not actually ran at all

* Migrations to be mssql compatible

Mssql driver doesn't handle if multiple queries are sent to same transaction concurrently.

* Prevented mssql failing when invalid schema builder was executed by accident

Instead of trying to generate sql from broken schema calls, just make exception to leak before query compiling is started

* Fixed mssql trx rollback to always throw an error

Also modified some connection test query to be mssql compatible

* Fixed various bugs from MSSQL driver to make tests run

* Fixed mssql unique index to be compatible with other dialect implementations

* Enable running mssql tests on CI

* Test for #2588

* Updated tests to not be dependend on tables left from previous test rans

* Trying to make mssql server work on travis

* Updated changelog and version

* Drop mariadb support  (#2681)

* Dropped support for mariasql

* ESLint

* Fixed docs to build again

* Fix knex.initialize() and adds test (#2477)

* Fix knex.initialize() and adds test

* Fix knex.initialize() and adds test

* Added test for reinitializing pool after destroy

* Fixed destroy / reinitialization test

* Fixed the fix

* Implement missing schema support for mssql dialect

* chore: Update dependencies. Remove estraverse (#2691)

* Update dependencies. Remove estraverse

* Fix compatibility with new Sinon

* Increase mssql timeout

* Normalized and validated driverNames of test db clients and fixed oracle test setup (#2692)

* Normalized and validated driverNames of test db clients and fixed oracle test setup

* Fixed failed queries from old query building tests which hadn't been ran in ages

* Allow selecting node version which is used to run oracledb docker tests

* Improved sql tester error messages

* Fixed rest of the oracledb tests

* Removed invalid flag from docker-compose

* Print mssql logs if initialization fails

* Fixed syntax error + final tests

* Added restart of failure for mssql DB initialization to try again if server was not ready

* Printout always mssql logs after container is started

* Fixed wait time printing after trying to connect

* Use npm run oracledb:test for testing oracle in travis

* Add Prettier (#2697)

* Add prettier
* Run files through prettier

* Istanbul -> NYC (#2700)

* istanbul -> nyc

* Update mocha

* Enforce code coverage (#2702)

* Enforce code coverage
* Update coverage numbers with current values

* version assignment on base class, copy on tx client, fix #2705

* Update changelog

* v0.15.1

* Added build step to test script. Fixes #2541 (#2712)

* Revert "Added build step to test script. Fixes #2541 (#2712)" (#2714)

This reverts commit 90ed8db.

* Allow oracle failures for now

* Fix issue with select(0) (#2711)

* Fix issue with select(0). Fixes #2658

* Refactor migrator (#2695)

* Refactor migrator

* Fix exports

* Fix ESLint

* Fix migrator

* Fix reference to config

* Split some more

* Fix table builder

* Fix argument order

* Merge branch 'master' of https://github.com/tgriesser/knex into feature/2690-support-multiple-migration-dirs

# Conflicts:
#	src/migrate/index.js
#	test/index.js
#	test/unit/migrate/migrator.js

* Fix #2715 (#2721)

* Fix #2715, explicitly set precision in datetime & timestamp
* Allow for precision in knex.fn.now, mysql time
* Add test for datetime with precision

* Bump changelog

* 0.15.2

* Fix issues with warnPromise when migration does not return a promise. Fixes #2725 (#2730)

* Add tests for multiple union arguments with callbacks and builders (#2749)

* Add tests for multiple union arguments

* Add some callback and raw tests to union unit tests

* Compile with before update so that bindings are put in correct order (#2733)

* Fix join using builder withSchema. (#2744)

* Use Datetime2 for MSSQL datetime + timestamp types (#2757)

* Use Datetime2 for MSSQL datetime + timestamp types

Datetime2 is now the recommended column type for new date work: https://docs.microsoft.com/en-us/sql/t-sql/data-types/datetime-transact-sql?view=sql-server-2017

* Add tests for MSSQL datetime2 changes

* General/document breaking change (#2774)

* Add changelog entry for a breaking change

* Improve entry

* Allow timestamp with timezone on mssql databases (#2724)

* Allow timestamp with timezone on mssql databases

* Change timestamp parameter to use object instead of boolean

* Update dependencies (#2772)

* Feature/2690: Multiple migration directories (#2735)

* Implement support for multiple migration directories

* Add tests for new functionality

* Fix tape tests

* Pass migration directory upwards

* Fix multiple directory support

* Fix bugs

* Rollback correctly

* Fix remaining tests

* Address comments

* #2758: Implement fail-fast logic for dialect resolution (#2776)

* Implement fail-fast logic for dialect resolution, clean-up code around.

* Remove method that was deprecated long time ago

* Address additional comments

* Try addressing comments

* Set client explicitly

* Fix compatibility with older Node versions

* Use columnize instead of wrap in using(). (#2713)

* Use columnize instead of wrap in using().

This is an attempt to fix #2136. Also added an integration test, couldn't find any existing.

* Exclude MSSQL from test as it does not support .using

* Change test to not use subquery

* Change test

* Introduced abstraction for getting migrations (#2775)

* Introduced abstraction for getting migrations

This would allow a webpack migration source which is compatible with bundling.

* Fixed migration validation with custom migration source

* Fixed issues after rebasing on muti directory PR

* Renamed parameter and fixed error message

* Addressed some PR comments

* Finished comment

* Moved filename extension filtering into fs-migrator

* Added test showing in memory custom migration source

* Cleaned up how to get config

* Fixed failing test

* Hopefully fix tests

* Fix Node.js 10 support in tests

* Test for correctly releasing cancel query connection
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants