seth/rails-7.2 by snickell · Pull Request #60437 · code-dot-org/code-dot-org

snickell · 2024-08-15T06:57:31Z

Follow-on to the Rails 3.3.4 upgrade PR, this PR upgrades Rails to 7.2.

Read changelogs/git-commits of Rails-related gems, and upgrade those who's changelog mentions changes to support Rails 7.x.
bundle update rails .... keep adding to the list until no Gem conflicts.
Disable composite_primary_keys gem, as support is now present in Rails as of 7.1. See: https://www.honeybadger.io/blog/composite-keys-in-rails/
Disable mysql_check_index_used.rb, at least temporarily, see: related comment

TODO

application.rb: # config.action_mailer.delivery_job = 'MailDeliveryJob', see seth/rails-7.2 #60437 (comment)
initializers/backtrace_silencers.rb: remove, see seth/rails-7.2 #60437 (comment)
Check how rails 7's built-in composite primary key support works with the annotate gem (we are using the latest version, this may change some model headers, it doesn't mark primary composite keys as primary in each item).
Decide if we want to remove mysql_check_index_used.rb, see: related comment

…es selenium-webdriver to 4.10.0

… anyway) which was spewing warnings to console

… - remove when bundler version is >=2.0'

…undle version (bundle does this automatically to match Gemfile.lock bundled with on bundle install)

…m to 3.5.17 Gem is an interesting case, the default gem version with ruby with 3.3.4 is 3.5.11, but a previous update of default.rb by Elijah had trouble unless a version of rubygem was specified that was NOT the system version, see: bf41ebf

…bled = no verification) cdo-ruby/ ruby_spec.rb: verify versions

…ws about Ruby 3.3.4.

…r chef will ignore if files already exist from a previous version

…lly use

…com/rbenv/ruby-build/wiki#ubuntudebianmint => build not broken

… "000000" were turning into the int 0 which confuses the poor badger

… but will no longer be part of the default gems since Ruby 3.4.0"

….org/issues/38118

…river

…-kitchen 3.4.0. See: #60329 (comment)

snickell · 2024-08-22T14:42:22Z

/Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/attribute_methods/serialization.rb:183:in `serialize': wrong number of arguments (given 2, expected 1) (ArgumentError)
        from /Users/seth/src/code-dot-org/dashboard/app/models/sections/section.rb:96:in `<class:Section>'
        from /Users/seth/src/code-dot-org/dashboard/app/models/sections/section.rb:42:in `<top (required)>'

Fix is to explicitly pass coder: to serialize as a keyword arg.

Fixed by: d33d7e9

…e primaries

snickell · 2024-08-22T14:51:20Z

migrating off composite_primary_key to rails built-in

/Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/dynamic_matchers.rb:22:in `method_missing': undefined method `primary_keys=' for class DatablockStorageTable (NoMethodError)
        from /Users/seth/src/code-dot-org/dashboard/app/models/datablock_storage_table.rb:32:in `<class:DatablockStorageTable>'

Fixed by: 7a8d698

snickell · 2024-08-22T15:05:59Z

/Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/mocha-1.1.0/lib/mocha/integration/mini_test/adapter.rb:27:in `included': uninitialized constant MiniTest (NameError)

          Mocha::ExpectationErrorFactory.exception_class = ::MiniTest::Assertion
                                                           ^^^^^^^^^^
Did you mean?  Minitest
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/mocha-1.1.0/lib/mocha/integration/mini_test.rb:43:in `include'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/mocha-1.1.0/lib/mocha/integration/mini_test.rb:43:in `activate'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/mocha-1.1.0/lib/mocha/mini_test.rb:3:in `<main>'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/3.3.0/bundled_gems.rb:74:in `require'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/3.3.0/bundled_gems.rb:74:in `block (2 levels) in replace_require'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/zeitwerk-2.6.17/lib/zeitwerk/kernel.rb:34:in `require'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/bootsnap-1.18.4/lib/bootsnap/load_path_cache/core_ext/kernel_require.rb:30:in `require'
        from /Users/seth/src/code-dot-org/dashboard/test/test_helper.rb:25:in `<main>'

Closely related issue: freerange/mocha#614

This could be fixed by upgrading mocha, but its a major version jump and seems like it could open a can of worms. Instead I've opted to work around this by pulling in the old MiniTest constants using require 'minitest/unit' which contains a compat layer.

Upgrading mocha+minitest could be revisited in a follow-up PR, probably a good idea anyway.

Fixed by: 3e7344a

snickell · 2024-08-22T15:06:29Z

bundled_gems.rb:74:in `require': cannot load such file -- i18n_controller (LoadError)
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/3.3.0/bundled_gems.rb:74:in `block (2 levels) in replace_require'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/zeitwerk-2.6.17/lib/zeitwerk/kernel.rb:34:in `require'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/bootsnap-1.18.4/lib/bootsnap/load_path_cache/core_ext/kernel_require.rb:27:in `require'
        from /Users/seth/src/code-dot-org/dashboard/test/controllers/i18n_controller_test.rb:2:in `<main>'

Another require that should be loaded by zeitwerk automatically.

Fixed by: 234a3b3

snickell · 2024-08-22T15:14:03Z

Remove lock_thread.rb, a "feature backport from Rails 5.1"

code-dot-org/dashboard/test/testing/lock_thread.rb:25:in `with_connection': undefined method `[]' for nil (NoMethodError)

    unless (conn = @thread_cached_conns[owner_thread])
                                       ^^^^^^^^^^^^^^
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/connection_handling.rb:296:in `with_connection'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/relation/calculations.rb:317:in `block in pluck'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/relation.rb:1452:in `skip_query_cache_if_necessary'
        from /Users/seth/.rbenv/versions/3.3.4/lib/ruby/gems/3.3.0/gems/activerecord-7.2.0/lib/active_record/relation/calculations.rb:313:in `pluck'
        from /Users/seth/src/code-dot-org/dashboard/app/models/unit.rb:861:in `unit_names_by_curriculum_umbrella'

This was fixed in rails here: rails/rails#28083

We backported the feature, but the backport no longer works and would appear to be redundant in modern rails.

If all the tests break in crazy non-deterministic ways, this might be a good comment to revisit ;-)

Removed by: 6dbd485

Fix transactional_test_case.rb

ERROR["teardown_all", Pd::InternationalOptInControllerTest, 7.714712999761105]
 teardown_all#Pd::InternationalOptInControllerTest (7.71s)
Minitest::UnexpectedError:         NoMethodError: undefined method `lock_thread=' for an instance of ActiveRecord::ConnectionAdapters::ConnectionPool
            test/testing/transactional_test_case.rb:41:in `block (3 levels) in <module:TransactionalTestCase>'
            test/testing/transactional_test_case.rb:39:in `each'
            test/testing/transactional_test_case.rb:39:in `block (2 levels) in <module:TransactionalTestCase>'
            test/testing/setup_all_and_teardown_all.rb:54:in `block (2 levels) in run_callbacks'
            test/testing/setup_all_and_teardown_all.rb:53:in `block in run_callbacks'
            test/testing/setup_all_and_teardown_all.rb:52:in `run_callbacks'
            test/testing/setup_all_and_teardown_all.rb:37:in `run'

Rails 7.2 removes def lock_thread= from connection_pool.rb in rails/rails#50999, replacing it with def pin_connection! and def unpin_connection!, which ALSO manage the transaction creation/rollback, fortunately in a way that exactly mirrors what we have in transactional_test_case.rb. As a result, I've replaced lock_thread= with a call to pin_connection!, as well as opening and rolling back the transaction.

Fixed by: de3a5f8

…nt and broken

snickell · 2024-08-22T15:22:49Z

Tests now start to run, but ALL fail with errors exactly like:

ERROR["test_deleted_user_cannot_sign_in", "SessionsControllerTest", 0.0067719994112849236]
 test_deleted_user_cannot_sign_in#SessionsControllerTest (0.01s)
Minitest::UnexpectedError:         NoMethodError: undefined method `fetch' for nil
            test/testing/setup_all_and_teardown_all.rb:36:in `run'

Repros with: bundle exec spring testunit ./test/models/lesson_activity_test.rb

Well this was a crazy one and took almost an hour to find. This bizarre error stems from not having a config/cable.yml file. ActionCable::TestCase is in the inheritance hierarchy for TestCase now, and it must try to load this file in its init. Solution is to paste in a commented out copy of the stock config/cable.yml from a fresh Rails 7.2 app. Hopefully we'll be turning this on soon, but not today ;-)

Fixed by: 5243242

snickell · 2024-08-22T16:10:33Z

A bunch of LevelsHelperTest based tests failing with this backtrace:

ERROR["test_app_options_sets_level_requires_channel_to_true_if_level_is_channel_backed", "LevelsHelperTest", 1.076027000322938]
 test_app_options_sets_level_requires_channel_to_true_if_level_is_channel_backed#LevelsHelperTest (1.08s)
Minitest::UnexpectedError:         ArgumentError: unknown keyword: :database
            app/helpers/read_replica_helper.rb:21:in `connected_to'
            app/models/channel_token.rb:47:in `find_or_create_channel_token'
            app/helpers/levels_helper.rb:99:in `get_channel_for'
            app/helpers/levels_helper.rb:213:in `app_options'
            test/helpers/levels_helper_test.rb:308:in `block in <class:LevelsHelperTest>'
            test/testing/setup_all_and_teardown_all.rb:36:in `run'

Repros locally with: bundle exec spring testunit ./test/helpers/levels_helper_test.rb

In Rails 6.0.1 connected_to has this signature:
connected_to(database: nil, role: nil, prevent_writes: false, &blk)

In Rails 7.2 connected_to has this signature:
connected_to(role: nil, shard: nil, prevent_writes: false, &blk)

Fix is to modify connected_to in read_replica_helper.rb to neither take nor pass on a database: keyword arg.

Fixed by: 0faac07

…arg in Rails 7.2

…lock_thread= (removed in Rails 7.2) See: rails/rails#50999

snickell · 2024-08-22T17:28:01Z

WARN: Multiple ActiveRecord connections are in use

Still getting lots of warnings:

WARN: Multiple ActiveRecord connections are in use, this can make transactional tests fail in weird ways.

Repros with: bundle exec spring testunit ./test/controllers/api/v1/pd/teacher_attendance_report_controller_test.rb

in transactional_test_case.rb we see:

        setup do
          pools = ActiveRecord::Base.connection_handler.connection_pool_list
          all_connections = pools.map(&:connections).flatten
          if all_connections.many?
            warning = 'WARN: Multiple ActiveRecord connections are in use, this can make transactional tests fail in weird ways.'
            CDO.log.warn warning
            puts warning
        end

Rails 7.2

In-situ

If we inspect all_connections and pools in-situ with binding.irb:

[test] dashboard > all_connections
=> [#<ActiveRecord::ConnectionAdapters::Mysql2Adapter:0x00000000134d18 env_name="test" role=:writing>, 
#<ActiveRecord::ConnectionAdapters::Mysql2Adapter:0x00000000134d18 env_name="test" role=:writing>]
[test] dashboard > ActiveRecord::Base.connection_handler.connection_pool_list
=> [#<ActiveRecord::ConnectionAdapters::ConnectionPool env_name="test" role=:writing>, 
#<ActiveRecord::ConnectionAdapters::ConnectionPool env_name="test" role=:writing>]

In other words, we have multiple connections because ActiveRecord::Base.connection_handler.connection_pool_list is returning two pools, both with the writing role connection in them.

rails console

What's interesting is that if we load rails console, we see more what we'd expect: one reading pool and one writing pool:

[test] dashboard > ActiveRecord::Base.connection_handler.connection_pool_list
=> 
[#<ActiveRecord::ConnectionAdapters::ConnectionPool env_name="test" role=:writing>,
 #<ActiveRecord::ConnectionAdapters::ConnectionPool env_name="test" role=:reading>]

Rails 6.1.7.7

in-situ

[test] dashboard > pools.length
=> 1

rails console

[test] dashboard > ActiveRecord::Base.connection_handler.connection_pool_list.length
=> 1

Fix

We hardcode our read/write replicas (following the pattern from a medium article) in active_record_roles.rb and embedded in application.rb like so:

  connects_to database: {
    writing: Policies::ActiveRecordRoles.get_writing_role_name, # => :primary in test
    reading: Policies::ActiveRecordRoles.get_reading_role_name # => also :primary in test
  }

Even when these are both the same connection pool, Rails 7.2 lists the connection pool twice. As a result, we trigger the warning. The fix is to de-duplicate our pools list. This is safe because we lock at the pool level.

Fixed by: 619a8fc

…tries

snickell · 2024-08-22T17:48:02Z

alias_attribute doesn't work for associations anymore

ERROR["test_not_logged-in_user_calling_get_show_by_keys_should_receive_success_to_", "ProgrammingClassesControllerTest::ProgrammingClassesControllerAccessTests", 202.97481100074947]
 test_not_logged-in_user_calling_get_show_by_keys_should_receive_success_to_#ProgrammingClassesControllerTest::ProgrammingClassesControllerAccessTests (202.97s)
Minitest::UnexpectedError:         NameError: undefined local variable or method `categories' for an instance of ProgrammingEnvironment
            app/models/programming_environment.rb:146:in `block in categories_for_navigation'
            app/models/programming_environment.rb:145:in `categories_for_navigation'
            app/controllers/programming_classes_controller.rb:89:in `show_by_keys'
            app/controllers/application_controller.rb:187:in `with_locale'
            test/test_helper.rb:506:in `block (2 levels) in test_user_gets_response_for'
            test/testing/capture_queries.rb:27:in `assert_queries'
            test/test_helper.rb:505:in `block in test_user_gets_response_for'
            test/testing/setup_all_and_teardown_all.rb:36:in `run'

It will only work for attributes, this was deprecated in Rails 7.1, but actually dropped in Rails 7.2. We aren't the only ones feeling some pain from this, see stackoverflow: aliasing associations in rails 7.1 for example. Shopify opened a PR that added an alias_association feature, which is still open, but did not get merged into Rails 7.2 unfortunately: rails/rails#49801

Workarounds

Use alias_method. This should work for read-only associations, but will not work for property= association methods.
A related rails issue (has_many through alias_method rails/rails#51062) suggests a workaround of adding multiple associations that point to the same table.
Avoid using the alias: in some cases we may only use one or the other other of the aliased attribute. In this case, removing the unused association is an obvious fix that simplifies code.

Fixing

A quick search suggests we use alias_attribute in 10 places.

5 of these places its pretty clearly an actual attribute being aliased, and I think nothing needs changing:

dashboard/app/models/backpack.rb:
  20    # to reflect the new table name, so an alias is used to clarify which table this ID maps to.
  21:   alias_attribute :project_id, :storage_app_id
dashboard/app/models/channel_token.rb:
  30:   alias_attribute :project_id, :storage_app_id
dashboard/app/models/featured_project.rb:
  22:   alias_attribute :project_id, :storage_app_id
dashboard/app/models/project_commit.rb:
  20:   alias_attribute :project_id, :storage_app_id
dashboard/app/models/course_version.rb:
  55:   alias_attribute :version_year, :key

For this one, I dropped :programming_environment_categories, which was afaict unused in favor of its alias categories:

dashboard/app/models/programming_environment.rb:
  26:   alias_attribute :categories, :programming_environment_categories

After change, related tests still pass: bundle exec spring testunit ./test/models/programming_environment_test.rb

Fixed by: 4c94df7

As far as I can tell RegionalPartnerCohort is actually an unused vestigial table. I'm following up with acquisition products to confirm this. Nevertheless, I ported it, following the "duplicate the association" pattern, in case there's some stealth use of it I can't confirm this is the safest pattern

dashboard/app/models/pd/regional_partner_cohort.rb:
  39:   alias_attribute :members, :users

After change related tests pass: bundle exec spring testunit ./test/models/pd/regional_partner_cohort_test.rb

Fixed by: 3486912

Update: @dmcavoy confirms that this is probably a vestigial table that we can (should!) drop

Here I discovered that two of the aliases were real attributes. I wasn't sure what qualified, with grades having a serializer, but I tested it and it works fine. For alias_attribute :teacher, :user I wasn't able to convince myself that one or the other could be reliably switched to because the words are generic and hard to search for. Thus I followed the recommended pattern of just duplicating the association.

dashboard/app/models/sections/section.rb:
  66:   alias_attribute :teacher, :user 
  92:   alias_attribute :lesson_extras, :stage_extras # this looks like a real attribute
  98:   alias_attribute :grades, :grade # I tested this and its a real attribute

Fixed by: 5f4f7fd

…uplicating association

…iation

snickell added 30 commits August 9, 2024 23:12

ruby => 3.3.4

4ea2cac

Gemfile.lock: bundle update cld, cld => 0.13.0

b8efdac

Gemfile.lock: bundle update css_parser, File.exists? no longer exists

3b08a92

yaml.rb: YAML.load only takes one arg with psych

c2afbb1

Gemfile: eyes_selenium => 6.3.7 + bundle update eyes_selenium updat…

98a6fca

…es selenium-webdriver to 4.10.0

bin/dashboard-server loads studio

2a85afd

remove our custom fork of sprockets (backport of rails 4=>3, outdated…

e37d4ea

… anyway) which was spewing warnings to console

pycall => 1.5.2

cdb9cda

ChatClient: print messages in dev too

346468a

Remove workaround from Gemfile marked 'This is a temporary workaround…

30f28ed

… - remove when bundler version is >=2.0'

Gemfile.lock: bundle => 2.5.17

ca474c4

Remove unnecessary (?) gem install bundler step from SETUP.md

fca7ba2

Dockerfile: ruby => 3.3.4, and remove unnecessary manual install of b…

13e355f

…undle version (bundle does this automatically to match Gemfile.lock bundled with on bundle install)

.drone.yml: use codedotorg/code-dot-org:ruby-3.3.4

998f918

Redo rake => 13.2.1 upgrade in /Gemfile

fa47a9e

cdo-ruby/.kitchen.yml: use ubuntu-20.04

ade3f94

cdo-ruby/.kitchen.yml: enable serverspec verifier (currently was disa…

fc84ba9

…bled = no verification) cdo-ruby/ ruby_spec.rb: verify versions

cdo-ruby ruby-build.rb: bump RUBY_BUILD_VERSION to 20240727 so it kno…

f07b5ec

…ws about Ruby 3.3.4.

cdo-ruby ruby-build.rb: include RUBY_BUILD_VERSION in tarball names o…

6c9611a

…r chef will ignore if files already exist from a previous version

cdo-ruby ruby-build: switch from serverspec to inspec which we genera…

8924b8a

…lly use

cdo-ruby ruby-build.rb: install apt-package list from https://github.…

99cc85c

…com/rbenv/ruby-build/wiki#ubuntudebianmint => build not broken

ruby_spec.rb: convert from serverspec to inspec

168acbd

honeybadger.yml: ensure API key is a string not an int, dev values of…

2568561

… "000000" were turning into the int 0 which confuses the poor badger

Fix pre-Ruby-3.4 warnings: "___ was loaded from the standard library,…

00f4bc1

… but will no longer be part of the default gems since Ruby 3.4.0"

Gemfile: redcarpet =>3.6.0 to fix a warning, see: https://www.redmine…

fddc4b0

….org/issues/38118

cookbooks/.ruby-version: bump to 3.3.4

5b2cf11

bundle update docker-api excon to fix kitchen create for dokken d…

9621c8b

…river

cookbooks/Gemfile: explicitly require racc, an undeclared dep of test…

445b360

…-kitchen 3.4.0. See: #60329 (comment)

cookbooks/Gemfile.lock: bundle update berkshelf

d9d9569

snickell added 3 commits August 22, 2024 04:20

LTI: remove requires, add namespacing to dashboard/lib/clients

209d8ac

user_event.rb: remove rails 8 enum deprecation warnings

d186c0b

section.rb: serialize using coder: GradesArray

d33d7e9

snickell added 2 commits August 22, 2024 04:45

Fix "Defining enums with keyword arguments is deprecated" warnings

4e6de6c

datablock_storage: self.primary_keys => self.primary_key for composit…

7a8d698

…e primaries

snickell added 2 commits August 22, 2024 04:55

lti_v1_controller: remove redundant (?) requires that make zeitwerk mad

62db1b7

test_helper.rb: require 'minitest/unit'

3e7344a

i18n_controller_test.rb: use zeitwerk not manual require

234a3b3

lock_thread.rb: remove code backported from rails 5.1 and now redunda…

6dbd485

…nt and broken

snickell force-pushed the seth/rails-7.2 branch from a596075 to 6dbd485 Compare August 22, 2024 15:15

LTI: remove more redundant requires

ee2f0c7

snickell added 2 commits August 22, 2024 06:03

config/cable.yml: add to fix undefined method fetch' for nil` on tests

5243242

cable.yml: explain why we had to create this file

32722a2

snickell added 2 commits August 22, 2024 06:23

read_replica_helper.rb: connected_to doesn't take a database keyword …

0faac07

…arg in Rails 7.2

transactional_test_case.rb: use pool.pin_connection! instead of pool.…

de3a5f8

…lock_thread= (removed in Rails 7.2) See: rails/rails#50999

transactional_test_case.rb: don't warn when there's duplicate pool en…

619a8fc

…tries

snickell added 3 commits August 22, 2024 08:52

programming_environment.rb: no alias_attribute on associations

4c94df7

regional_partner_cohort.rb: no alias_attribute associations, fix by d…

3486912

…uplicating association

section.rb: no alias_attribute associations, fix by duplicating assoc…

5f4f7fd

…iation

snickell force-pushed the seth/rails-7.2 branch from 0e498ce to 5f4f7fd Compare August 22, 2024 19:42

Base automatically changed from seth/ruby-3.3 to staging January 9, 2025 03:06

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

seth/rails-7.2#60437

seth/rails-7.2#60437
snickell wants to merge 157 commits into
stagingfrom
seth/rails-7.2

snickell commented Aug 15, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024

Uh oh!

snickell commented Aug 22, 2024

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

snickell commented Aug 22, 2024 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

snickell commented Aug 15, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

TODO

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snickell commented Aug 22, 2024

migrating off composite_primary_key to rails built-in

Uh oh!

snickell commented Aug 22, 2024

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Remove lock_thread.rb, a "feature backport from Rails 5.1"

Fix transactional_test_case.rb

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

WARN: Multiple ActiveRecord connections are in use

Rails 7.2

In-situ

rails console

Rails 6.1.7.7

in-situ

rails console

Fix

Uh oh!

snickell commented Aug 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

alias_attribute doesn't work for associations anymore

Workarounds

Fixing

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

snickell commented Aug 15, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading

snickell commented Aug 22, 2024 •

edited

Loading