* Add dual booting logic to Gemfile
This might be helpful for the Rails 6.0 upgrade project.
* Add more than one gemfile to Travis' configuration
We want to see how the application behaves with more than one Rails versions:
- Gemfile -> Rails 5.2
- Gemfile.next -> Rails 6.0
This will help us figure out what needs to be addressed before migrating to Rails 6.0.
If you want to read more about this technique (dual booting) you can check out this page: https://www.fastruby.io/blog/upgrade-rails/dual-boot/dual-boot-with-rails-6-0-beta.html
* Fix joins
* Upgrade Gemfile.next.lock
* Make sure we're installing the correct versions of gems
* Add Rails 6 notes
* Update rubocop in Gemfile.next.lock
* Fix organization spec
* Fix page_views_spec
* Add Rails 6 and run rails app:update
* Remove some tricks
* Remove Gemfile.next for now
* Fix .content_type deprecation
* Fix deprecation of .where.not NAND/NOR behavior
* Fix deprecation of parameterized emails
* Fix specs
* Remove next flag for now
* Fix spec (hopefully)
* Add wait_for_javascript
* Fix spec, thanks @maestromac!
* Try without wait for javascript hack
* Remove unnecessary bin/update
* Remove file that snuck in the rebase
* Update the vendored gems
* Replace migrate+db:setup with db:prepare
* Update vendored gems
* Fix Gemfile.lock and update vendored stuff
* Fix Gemfile.lock to be the same as master's minus the changes
Co-authored-by: rhymes <rhymesete@gmail.com>
* Add 'top_comments' to article feed
* Only include root comments
* Cache and eagerly load top comments
* Clear top comments cache when article score is updated
* Add articles.top_comments to bullet whitelist
* Update app/models/article.rb
Co-Authored-By: rhymes <rhymesete@gmail.com>
* Remove limit in top_comments association, remove Rails caching of top comments
* Remove unnecessary test
Co-authored-by: rhymes <rhymesete@gmail.com>
* Upgrade to Ruby 2.7.0
* Upgrade Travis's Ruby as well
* Fix conflict
* Silence all Ruby 2.7 deprecation warnings in dev mode
* Silence dev and test unless CI
* Add more Broadcasts to seeds for welcome notification flow
* Remove unnecessary welcoming_user_id from development config
* Add basic service class for generating appropriate broadcast per user
Eventually, this should be called from a rake task that will rely on this service to determine which broadcast, if any, is the appropriate one to send to newly signed-up users, and will enqueue a worker to create/send a welcome notification when appropriate.
* Add a few TODOs, plus some general cleanup
* Add some skipped tests for WelcomeNotification::Generator
* Add "Welcome" type to Broadcasts, allow dynamic "authoring" of Broadcasts
This adds a new `type_of` to the Broadcast model: "Welcome". As we begin to create a new "welcome notification" workflow, we'll categorize them by making them all of the same type.
This also adds a concept of a "welcoming user", which can be set via an ENV var. The WELCOMING_USER_ID will allow us to explicitly set (and change) which user in the database is the one that "sends" welcoming notification. In production, we plan for this to be dev.to/sloan.
* Allow WelcomeNotificationWorker to accept + send any kind of broadcast
This abstracts out the logic of deciding which broadcast to send from the woker into the calling method. This will help us send many different kinds of welcome notifications using one, resuable worker class.
* Raise if a Notification can't be created
* Add STAFF_USER_ID as an ENV var
This is in preparation for moving away from SiteConfig.staff_user_id, and replacing it with an ENV var. We will need to set this in production first, and then make a separate PR to replace all instances of staff_user_id with the newly-set ENV var.
* Fix logging for dev and test environments
We had some unused code for Timber in our config files for development and test.
However, this was completely clobbering Rails' default behavior of logging out to log/development.log and log/test.log. We should allow for logging in these environments in particular, and since we're not using Timber there, we can just revert to using the Rails default debugging log levels.
* Add Timber logger to specs that explicitly require it
* Restore default Rails.logger in specs using Timber logger
* Explicitly use Rails.logger in OmniauthCallbacksController
We seem to have different logger classes leaking into this controller in tests.
I'm not entirely sure why this is/haven't been able to investigate yet, but I do think we should be using the Rails.logger explicitly until we can figure that out.
`redis_store` was initially introduced to switch sessions from the cookie store to Redis in https://github.com/thepracticaldev/dev.to/pull/4004. When we introduced a dual cache store in https://github.com/thepracticaldev/dev.to/pull/4991 we forgot to use the builtin Rails cache store - [ActiveSupport::Cache::RedisCacheStore](https://guides.rubyonrails.org/caching_with_rails.html#activesupport-cache-rediscachestore) - and instead kept using the external [ActiveSupport::Cache::RedisStore] which comes from the `redis-activesupport` gem included by `redis-rails`.
The gem `redis-activesupport` contains this disclaimer:
> Rails 5.2.0 includes a Redis cache store out of the box, so you don't really need this anymore if you're generating a new Rails application. We are no longer accepting new features for this gem, only pull requests for security and compatibility fixes will be accepted.
Other reasons to use the Rails builtin cache store, other than being supported and developed are the following:
- supports and uses `hiredis` as a client if installed:
22483b86a6/activesupport/lib/active_support/cache/redis_cache_store.rb (L12-L14)
- supports compressions for keys over 1kb (that can be configured)
- is evolved with Rails itself
* Initial work for streaming app shell way of using serviceworkers
* Close in on serviceworker adjustment finalization
* Add docs and loading indicator
* Remove useless code
* Don't run on API
* Update app/javascript/contentDisplayPolicy/hideBlockedContent.js
* Fix small details
* Don't run serviceworker.js code in test env
* Fix JS in serviceworker
* Move private keyword to proper place in async controller
* Change shell_version to HEROKU_SLUG_COMMIT
* Update caching config
* Add test for async_info/shell_version
* [WIP] Extract DEV specifics to Env variables
* [WIP] Fix sendgrid url that I accidentally changed. Womp.
* Remove extra colon from 'variable' calls
* Trim :// off of app protocol
* Update Envfile TIMBER_LOGS default comment for accuracy
* Add API doc links for each service
* Start with webhooks: table and model
* Start with webhooks api
* Start with webhooks api
* Webhook::Event class and events list
* Remove commented callbacks
* Start with sending events to webhook endpoints
* A couple of tests for Articles::Creator
* Send event to webhook endpoint on article destroy
* Dispatch event on article update
* Dispatch event when an article updated from admin
* Spec for the webhook job
* One more test for the dispatch event job
* Integration-like spec for event dispatching to webhook endpoints when creating an article
* Use Addressable::URI to parse endpoint url
* Add oj as a faster fast_jsonapi backend
* Renamed serializers
* Move article serialization out of a model
* Don't allow to create a webhook event with invalid type
* Add fields for ArticleSerializer
* Fix webhook event job specs
* Specify timeout when dispatching events
* Fix webhook job spec
* Change webhook events queue name
* Change serialized article fields for the dispatchable events
* Include user data when serializing an article for webhook event
* Moved decorating out of Webhook::DispatchEvent, fixed most specs
* Fix route in ArticleSerializer
* Move Article decoration to Webhook::PayloadAdapter
* Refactor image url helpers to avoid including helpers into serializer
* Fix specs
* Default url options for production
* Preload organization, not user
* Preload users only when needed
* Pre-load podcasts for podcast episodes in API
* Avoid eager loading error by loading rating votes separately
* Preload associations for moderation
* Preload user comments in trees
* Preload organization for non org dashboard and cleanup queries
* Optimize ArticleSuggester to only load N articles at need
* Remove eager loading and pass variables to partials for easier debug
* Reorganize tags validation code and ignore actsastaggableon eager loading issues
* Remove unused eager loading and bring up comments relation
* Preload podcasts when loading podcast episodes
* Fix views specs
* Make sure ArticleSuggester never returns duplicates
* Remove commented code
* Re-trigger build
* Move suggested articles back to view to respect fragment caching
