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

Version controlled seeder #215

Closed
wzrdtales opened this issue Feb 9, 2015 · 15 comments
Closed

Version controlled seeder #215

wzrdtales opened this issue Feb 9, 2015 · 15 comments
Assignees
Labels
Milestone

Comments

@wzrdtales
Copy link
Member

wzrdtales commented Feb 9, 2015

Implementation of a version controlled seeder, plus seeder in general.


Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

@wzrdtales
Copy link
Member Author

wzrdtales commented Jan 9, 2018

@BorntraegerMarc To answer on the state asked over in #292

No short answer here currently about the state.

There are some things I can do quickly and some that need a bit of love. Lets start with the things I can do quite quickly:

  • Push out seeders as is, the interface separatation is there basically just hooks would need to be added
  • Add some of the basic operations: insert, delete, update, search

What needs a bit of more love:

  • Make seeders actually useful and more than a separation of two completely different things (DML vs DDL)*
  • Rewrite the loader
  • Rewrite the seed and migrator logic
  • Separate file from seed and migrator logic
  • Separate walker from seed and migrator logic
  • Add version selector to choose correct seed and migrator logic
  • Add a mode switcher, to switch between seeds and migrations
  • Add a fail safe processor for large manipulations
  • To make them useful I would want to add some helpers that really help the user to pursue what he wants to do.

Thoughts? Would it be from benefit to get a seeder version quickly with just the very basic actions? Already thought about doing exactly this though, currently I do work to rewrite the whole logic that loads and processes the migrations and seeds as I look forward to have the new migration schema in as quickly as possible. Especially for seeds those will include important benefits, like atomic operations and some other goodies. If it would really benefit the community to have the rough implementation I might go ahead to push that out first, but that would mean that you could expect quite soon a change of how they work again. Which would not break in any as we already have support for multiple different versions of seed versions. A last option would also be do the rough version and release the seeders v1 version already with the new schema, but with hard limitations of features for them.

To add the fail safe processor was targeted anyway to be added later, but just listed it here for completeness.

@wzrdtales
Copy link
Member Author

wzrdtales commented Jan 9, 2018

To go a bit more into detail, mentioned goodies include:

A Concurrency manager - This will finally allow to safely integrate db-migrate into your application and just let it run when your application starts. Just one instance will operate at the same time. Basically I plan to insert a json metadata row as row 0 into the migrations and seed table (or maybe introducing a new table not sure yet, but that would break with the drivers, so I would like to avoid that)

an event emitter that emits events to let your application know when changes successfully processed which would allow to roll out changes very dynamically. I.e. you've got a new feature that depends on the existence of that one particular column, but does not care about the rest of the migrations, your application could then use feature toggles and toggle those with the help of those events.

Real atomic operations that can be rolled back in any scenario if wanted. Even without the help of transactions. Basically db-migrate would by default rely on transactions, as those are way faster than safe guards we can implement. But if the driver tells db-migrate there are no transactions for this db, than db-migrate could fall back to those routines. Or for example in the case of MySQL transactions can't be used to rollback DDL, in MariaDB there is an open task for this to enable this, but that is open for years already. Also here we can easily rollback without the help of transactions. And this also is possible in the other direction, i.e. for cockroachdb, where DDL within a transaction can return as completed instantly, while cockroach might delay the real execution of those as cockroachdb does by default online DDL.

wzrdtales added a commit that referenced this issue Jan 9, 2018
@wzrdtales
Copy link
Member Author

With this changes I just pushed to this branch 6f40d6b, seeders basically could be used already as far as the core is concerned though. Naturally the drivers would need to get the functionality added, that is not yet there.

@BorntraegerMarc
Copy link
Contributor

Nice summary, thanks! My point of view: I'm eager to test the seeders but not necessarily depending on it. So I would be totally fine to test out seeders in the separate branch and wait to use it in my application once all features have been implemented. I think the community thinks the same.

But something that would make my life easier: can we remove those long warnings in the console insert is obsolete use version controlled seeders instead? If we decide that we're gonna wait to release seeders until they are fully ready :)

@BorntraegerMarc
Copy link
Contributor

Hi @wzrdtales just wanted to ask if you've considered my idea to disable the console warning 🙂

@wzrdtales
Copy link
Member Author

Sorry! Yes I do consider it, it is a fair deal!

@wzrdtales
Copy link
Member Author

I do appreciate your engagement @BorntraegerMarc, it is just always an up and down for me when it comes to time that I've got left :)

@BorntraegerMarc
Copy link
Contributor

@wzrdtales no problemo. I understand the feeling only to well :)

Could you hit us with a message in case you decide to release a new version with the logs disabled? (Would really appreciate it)

@wzrdtales
Copy link
Member Author

Will do @BorntraegerMarc , currently am writing down some of the new logic and also refactoring some of the files to use const and let instead of var now that support prior 4.0.0 is dropped :)

wzrdtales added a commit that referenced this issue Feb 2, 2018
  start of the port of the version 1 schema.

Refers to #215
@wzrdtales wzrdtales mentioned this issue Feb 2, 2018
9 tasks
wzrdtales added a commit that referenced this issue Feb 2, 2018
wzrdtales added a commit that referenced this issue Feb 2, 2018
start of the port of the version 1 schema.

Refers to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
wzrdtales added a commit that referenced this issue Feb 2, 2018
wzrdtales added a commit that referenced this issue Feb 2, 2018
start of the port of the version 1 schema.

Refers to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
wzrdtales added a commit that referenced this issue Feb 2, 2018
wzrdtales added a commit that referenced this issue Feb 2, 2018
Referst to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
@wzrdtales
Copy link
Member Author

Good news are, I had some time today and I mostly finished the rough stuff for the loader. it at least works already.

wzrdtales added a commit that referenced this issue Feb 3, 2018
This temporarily removes the note being displayed to the user
when an insert call is being made. This will be removed until
seeders really going to be introduced.

Refers to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
@wzrdtales
Copy link
Member Author

@BorntraegerMarc
Released

This includes the refactoring to remove the deprecation notice for now.

Also the following fixes flown in

https://github.com/db-migrate/node-db-migrate/blob/467bf2a7ae72a05805015f7595335d5efbacd3ce/CHANGELOG.md#0103-2018-02-03

wzrdtales added a commit that referenced this issue Mar 27, 2018
wzrdtales added a commit that referenced this issue Mar 27, 2018
start of the port of the version 1 schema.

Refers to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
wzrdtales added a commit that referenced this issue Mar 27, 2018
Referst to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
wzrdtales added a commit that referenced this issue Apr 10, 2018
wzrdtales added a commit that referenced this issue Apr 10, 2018
start of the port of the version 1 schema.

Refers to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
wzrdtales added a commit that referenced this issue Apr 10, 2018
Referst to #215

Signed-off-by: Tobias Gurtzick <[email protected]>
RandomSeeded added a commit to blend/node-db-migrate that referenced this issue May 29, 2018
* cs fixes

* add concepting to comments

* fix driver inclusion

* remove sqlite dependency, add sqlite to devdependencies and outsource driver

* remove unnecessary files

* remove mongodb dependency, outsource mongodb driver

* Bump sqlite3 package version to fix issue with node 4.0.0

* Bump sqlite3 package version to fix issue with node 4.0.0

* add node version 4.0, remove iojs allowed build failures

* remove mysql dependency and add to dev for testing purposes

* fix index_test tests

* change base test to extern base

* fix index test

* remove last driver dependencies

resolves db-migrate#227

* add pg to dev dependencies

* ?

* updated fixed pg driver

* add error message to executeDB and handle callback

* return error message if available or the object/message directly passed through

* fix some cs issues

* add missing nosql methods to interface

* do not strip private functions from driver

removed unnecessary entries from migrator interface
do not strip any `_function` unless it is explicitly defined via exclude

* change config to internal scope

* fix test for new internal config

* fix last details to move config i into internal scope

* remove debug message

* remove es5 from jshint rules, fix one cs

*  add rules to jshint

* fix some cs issues and adjust rules

* adjust some settings and add jscs already

* [cs] fixed some issues and do some cleanup

* update lab

* update dependency and check

* add 4.1 to travis ci

* change couple of templates

* update sql templates

* add undo seed function to api

* fix bug on up/down of seeders

* add reset method and limit down methods of seeds

* add dummy calls to linked seeders and migrations

* link to read the docs and remove readme documentation

* 0.10.0-beta.1

* remove the badge ending from the documentation url

* make entry in readme more clear to be also a user manual

* Extend startup with logging and failure handling on 0.9.x versions

Before this patch users may ran into a local installation older than 0.9.x, this patch
ensures that the version is compatible to be preffered over the global version. If the
local version is to old (<0.10.0) the global version will be used instead. If this is
the case a warn message gets displayed to the user.

Additionally added informational logging within the verbose mode if a local version gets
preferred.

solves db-migrate#317

* 0.10.0-beta.2

* add lts node.js version to travis tests

* Resolve driver from local projects

Fixes db-migrate#318

* use synchronized variant of resolve for drivers

* 0.10.0-beta.3

* fix sql templates to work with 0.10.x again

* ignore on init formatting

* 0.10.0-beta.4

* Support setting default environment via ENV variable.

* Support setting default environment via ENV variable.

* coffeescript template refactory

coffeescript template refactory

by default coffeescript always returns the last line/variable, so there is no need to use the ~return~ keyword, there is not need for ; also.

* add scope logging

* [bugfix] internal linking causes infinite loop

fixes db-migrate#329

* add scope logging

* [bugfix] internal linking causes infinite loop

fixes db-migrate#329

* coffeescript template refactory

coffeescript template refactory

by default coffeescript always returns the last line/variable, so there is no need to use the ~return~ keyword, there is not need for ; also.

* 0.10.0-beta.5

* remove debug message

* 0.10.0-beta.6

* allow configuration through options object

* adjust test

* add direct execution of seeders and migrator links

* Fix config file setting.

* update parse db-url to 0.3.0

Introduces mongodb uri support

* update some dependencies

* fix(migration):

A promise is not properly identified and returning an empty resolve ends up in
db-migrate hanging up forever.

Fixes db-migrate#343

* fix(api): callback not called on run

If db-mgirate is run via any run method (except directly from console), the callback was not available
to the migration functions anymore.

* fix(api): callback called twice

Revert the change to add the callback if it is undefined.

Fixes db-migrate#343

* Fix promise issue for down migrations.

* refactor(migration): Make check for promise more reliable

* fix(test): Stub MySQL connect method instead of calling the original

The MySQL driver now connects directly to the driver, instead on the first query. This
caused some trouble with the ssh tunnel test, this functionality was stubbed.

Fixes db-migrate#348

* chore(security): Bump dependency because of security vulnerability

* pull down package.json

* chore(dependencies): update and remove unneeded dependencies

* Update data_type.js

For consistency :-)

* chore(travis): change travis config to use modern compiler

* chore(travis): add node 5 to ci builds

* chore(travis): remove obsolete node 4.1

Keep 4.2 lts and node 4 in the build which is usually enough.

* Avoid warning in promise/callback bridge code

When Bluebird detects when a promise is created within a promise
handler, but is not returned, it [emits a warning][warn] 'a promise was
created in a handler but was not returned from it'.

This happens if the `callback()` function is implemented using Promises,
but uses `nodeify()` to make it callback compatible. By using nodeify
here, it handles the returned Promise (if any) and avoids the warning.

 [warn]: https://github.com/petkaantonov/bluebird/blob/master/docs/docs/warning-explanations.md

* Updated documentation link in README

Current documentation link points to a 404 page, updated it with a working version.

* chore(travis): add node 6

* fix(api): fix race condition on create migration

A callback was placed not inside but after the callback which it belongs to. This
resulted in a very rarely occuring race condition when creating migrations without
an existing migration directory.

Fixes db-migrate#376

* 0.10.0-beta.12

* chore(compatibility): update dependencies

* chore(dev): add .tern file to project

* refactor(log): move log into new extern lib

Refers to db-migrate#382

* refactor(datatype): move dataType into new extern lib

Refers to db-migrate#382

* refactor(remove): remove log and dataType artifacts

Refers to db-migrate#382

* refactor(util): remove util artifact

Resolves db-migrate#382

* detect real callback for several api functions

* feat(api): promisify all current api methods

* refactor(ssh): delay loading of tunnel-ssh and improve load time

This improves the loadtime of db-migrate by 47%. To be exact from ~22ms to ~15ms.

Resolves db-migrate#387

* refactor(modularize): require inflection from npm instead

* refactor(tests): added shadow test and rewrite api test

* chore(comments): add some notes to the api tests

* refactor(globals): remove several globals, this might break things

There are several things that are exposed as global. For example Migrator, Seeder, Class, or the async and dbm globals.

Most probably finally removing async and dbm which have been marked deprecated, will end up for some people in a breaking change. But we need to remove all those
globals, to not interfere with projects using db-migrate. DB-Migrate is not just a cli module anymore though. In this step we now also finally removed internals completely
from the API, and needed to adjust the onComplete callback to actually do this. This change could also lead to a breaking change and should thus be communicated

All in all this is a big clean up and also two not necessary functions were removed.

* chore(parallelize): parallelize non interferring api tests

* refactor(tests): add index tests and some configs

* refactor(tests): refactored util_tests

* refactor(tests): refactored migration tests

* refactor(tests): refactored config tests

* refactor(tests): refactored base tests

* refactor(tests): add create tests and remove sample shadows

* chore(ci,dep): update ci and dependencies

* chore(test): create Makefile for tests and edit ci config

Removing 0.10 and 0.12 from tests

* chore(ci): clean up ci config

* chore(dependencies): upgrade dotenv and pkginfo

* chore(devdep): update and remove some devDependencies

* use path.join (db-migrate#353)

refactor(template): use path.join as it is intended to be used

* fix(template): fix unnoticed error introduced in the last merge request

Wrong migrations have been generated

* feat(plugins): add basic support for plugins and improve performance

Some libraries have been changed to be delay loaded. DB-Migrate is back to the minimal load time possible again, if no command at all is entered.

Basic plugin support and the first hook, in this case of the type overwrite has been added. An functional example of a yaml config plugin
has been published.
Also some old functions have been cleaned up which are not necessary any more.

refers to db-migrate#397

* chore(leftover): remove example plugin leftover from package.json

* fix(api): fix introduced undefined behavior of specified configs

internal.cwd added again, which is needed.

* fix(api): fix scoping

The api was missing the matching internal, which resulted in a missing prefix
when executing scope migrations.

fixes db-migrate#409

* most of the first protocol transition and the transition helper
is implemented already. Just the logic to actually call the helper is currently missing.

* feat(transitioner): add transitioner to easen the process of protocol changes

Resolves db-migrate#403

* feat(hook): parser hook and transitioner api

Added a hook for parsers and added the API endpoints for the transitioner.

Refers to db-migrate#403
Refers to db-migrate#397

* fix(config): Don't throw if environment variable is empty

Fixes db-migrate#411

* chore(hook): allow also to hook without adding any extension

* fix(transitioner): add new parser internal to transitioner

* fix(transitioner): catch whitespaces properly

* chore(templates): add meta protocol v1 to already aligned migration templates

* feat(plugin): add basic plugin support

Added basic plugin support, enabling to inject at multiple targets of the API, config and register
pre compiler.

Resolves db-migrate#397
Refers db-migrate#396

* fix(plugin): use correct path to include plugins

* fix(errorhandling): Add missing error assertion in executeDB

fixes db-migrate#381

* feat(sync): add sync mode

Refers to db-migrate#383
Resolves db-migrate#313
Refers to db-migrate#222

* refactor(tests): remove tests from main repo

Tests have been moved to their driver repo.

Resolves db-migrate#399

* refactor(cleanup): cleanup package.json and add db.config.ci back

* chore(ci): exclude interfaces

* chore(ci): Readd missing dependencies

* chore(test): Add Test for privateKey param on shh tunnel

* fix(resolve): Check if resolved version has plugin support

fix db-migrate#425

* log error on undefined stack

* refactor(api): cleanup information and added commentary

* feat(config): add rc style configs

fixes db-migrate#308
fixes db-migrate#406

* refactor(api): refactor tests and make api.js more modular

* fix(api): add missing reference to sync

* refactor(api): remove last output of postponed seeders

* add test dummies for api

* add first programable api tests

* directly use promises in tests

* Double quoting interpolated file name for coffee

* fix(create): use same timestamp in every created file

* fix(args): dont parse when called as module

fixes db-migrate#449

* fix deep config replacements

fixes db-migrate#473

* feat(config): helper to overwrite and extend configuration

This allows to parse a url from an url property and as well to overwrite and extend
variables as needed.
To pass in url as normal value an object is needed like the following: { url: { value: "http://example.com } }.

Fixes db-migrate#349
Fixes db-migrate/pg#8
Fixes db-migrate#488
Fixes db-migrate#463

* fix(tests): fix breaking tests

* refactor(config): do not just interpret database.json

also check for either a passed config object OR a config file exists

* fix(create): Fix create when using db-migrate as module

Bug introduced on db-migrate#485

Fixes: db-migrate#493

* test(create): Fix testing as module when no database.json on root

* fix small typo in cli help menu

* allow rc file to contain location of config file

* print the reason in unhandledRejection evnet if reason.stack does not exist

* complete refactor of api js

* add eslint to dependencies

* fix a few bugs and update tests

* fix last tests

* adjust to cs

* adjust manual changes for cs

* remove process.exit when no callback passed and remove optimist help when in module mode

Fixes db-migrate#516

* add changelog and release 0.10.0

* 0.10.0

* adjusted tests and updated devDependencies

Refers to db-migrate#518

* readd pkg-lock

* change tested versions

* rewrite latest to node

* add linting and fix cs on config test

* add commitlint

* remove parallel: true as it was deprecated

* add stale config

* use more appropiate label

* Create CODE_OF_CONDUCT.md

* fix(db): wrong reference to connect causes db:create to fail

Fixes db-migrate#520

* 0.10.1

* add changelog

* fix(log): error ended up in unreadable errors

Fixes db-migrate#524
Fixes db-migrate#521

* add changelog

* chore: Remove `moment`.

Signed-off-by: wtgtybhertgeghgtwtg <[email protected]>

* feat(issuetemplate): added a github issue template

Signed-off-by: BorntraegerMarc <[email protected]>

* fix(exitCode): wrong check for existence fixed

A wrong check for existence resulted in events not
being loaded on version sized like 4.0.0. This resulted
in non zero error codes when actually running into errors.

Signed-off-by: Tobias Gurtzick <[email protected]>

* refactor(deprecationNotice): remove insert deprecation

This temporarily removes the note being displayed to the user
when an insert call is being made. This will be removed until
seeders really going to be introduced.

Refers to db-migrate#215

Signed-off-by: Tobias Gurtzick <[email protected]>

* fix(exitCode): wrong exit code on db methods

DB Calls always returned an exit code of 0, which leads
to an unexpected behavior on the user side.

Fixes db-migrate#534

Signed-off-by: Tobias Gurtzick <[email protected]>

* fix(switchDatabase): no error was thrown on scope switch

switchDatabase error was unhandled, which resulted in an unhandled
scope switch error.

Fixes db-migrate#470

Signed-off-by: Tobias Gurtzick <[email protected]>

* 0.10.3

* add changelog

* fix(insert): add missing insert entry to interface

Fixes db-migrate#542

Signed-off-by: Tobias Gurtzick <[email protected]>

* 0.10.4

* changelog

Signed-off-by: Tobias Gurtzick <[email protected]>

* 0.10.5

* update to use real checkboxes

* Update ISSUE_TEMPLATE.md

* feat(contribution): enrich contribution instructions

Refers to db-migrate#549

Signed-off-by: Tobias Gurtzick <[email protected]>

* feat(contribution): enrich contribution instructions, issues

Signed-off-by: Tobias Gurtzick <[email protected]>

* feat(progamableApi): CMD options can be passed programatically now

Signed-off-by: Marc Bornträger <[email protected]>

* fix(ci): add ignores for backported features

Signed-off-by: Tobias Gurtzick <[email protected]>

* chore: update dependencies

Signed-off-by: Tobias Gurtzick <[email protected]>

* feat(progamableApi): using const now

Signed-off-by: BorntraegerMarc <[email protected]>

* add changelog

Signed-off-by: Tobias Gurtzick <[email protected]>

* 0.10.6

* chore: add test to detect cmdOptions functionality

Refers to db-migrate#560
Refers to db-migrate#557

Signed-off-by: Tobias Gurtzick <[email protected]>

* fix(progamableApi): cmdOptions get passed into setDefaultArgv now

Signed-off-by: Marc Bornträger <[email protected]>

* 0.10.7

* add changelog

Signed-off-by: Tobias Gurtzick <[email protected]>

* chore: update dependency

Signed-off-by: Tobias Gurtzick <[email protected]>

* chore: update dependency

Signed-off-by: Tobias Gurtzick <[email protected]>

* feat(check): add check functionality to determine migrations to run

Signed-off-by: RandomSeeded <[email protected]>

* 0.11.0

* add changelog

Signed-off-by: Tobias Gurtzick <[email protected]>

* fix(reset): regression introduced in check functionality

db-migrate#552 introduced a regression and effectively disfunctioned the reset function, since it
did append just to the actionables within the run context.

It was moved into its own context.

Signed-off-by: Tobias Gurtzick <[email protected]>

* add changelog

Signed-off-by: Tobias Gurtzick <[email protected]>

* 0.11.1

* fix(check): fix check via API not passing results to the callback

Signed-off-by: RandomSeeded <[email protected]>

* save

* save

* updates

* cleanup
@stanislavt
Copy link

@wzrdtales Any updates?

@wzrdtales
Copy link
Member Author

I wont have any time before before june, concerning the seeders they are still on hold if you want to help working on them you're welcome though :)

@Ash-oi
Copy link

Ash-oi commented Oct 26, 2018

hey @wzrdtales love your work, just wondering how this feature is going?

@wzrdtales
Copy link
Member Author

closed in favour of redeclaration

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants