deckard 0.3.0 → 0.4.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: a0545ee7c006b54aaedd28b162ee6eb06647b0812bfbbdcb3b5846a78d6cdd68
4
- data.tar.gz: 9cf0a020aaeabaf9d659549b3045fe9ba2b4a758ee40106163a6aead95c7a778
3
+ metadata.gz: 9d5a064c82d88bfb7e1303ab4fcd7ab08936f78c78c9b27256a95fcb79e22520
4
+ data.tar.gz: 71320bd238bb13cae37b84fb3d6f2fadef423dd997b93e483514e8c163208be2
5
5
  SHA512:
6
- metadata.gz: 33e82c2fbba3856af1019a61bc1cf181e6f74383824895ab667d76aedd1994a04af5e69534c73f379ba2b64f333748319767290511f97eaefb13fbb75e674b83
7
- data.tar.gz: dacd253d2915316b2549f6ea4196354055f0abd4b504b14c9e6980ecab11f83d676a1868e81751c785ee813fd242bd256361daaae475c5078eda6c227364d0aa
6
+ metadata.gz: 165d0c5ad4bc499f3fcf8d7c54ae27cde30451c1ee4a9d55ad9f86884a163ff9104c44bcf58d64213489e75cde619f78b1145b07a894a9837711746c0e18280b
7
+ data.tar.gz: 4820249beabe3bb2c8bf61657cb34396b2a727a950d6b0850638105ad7a8dbbc4cc9f444d92affae1c9dc7efaeea11ecdf5a0820195723413fce8f21bc8901b2
data/CHANGELOG.md CHANGED
@@ -5,7 +5,34 @@ All notable changes to deckard are recorded here. The format follows
5
5
  [Semantic Versioning](https://semver.org/). Before 1.0, minor versions may
6
6
  change the stream format and the Ruby API.
7
7
 
8
- ## [0.3.0] - Unreleased
8
+ ## [Unreleased]
9
+
10
+ ## [0.4.1] - 2026-11-17
11
+
12
+ ### Changed
13
+
14
+ - `natural_key` is now `match_by`, and `natural_keys` is now
15
+ `match_each_model_by`. The names say what the declaration does: the
16
+ destination matches an existing row by those attributes. There are no
17
+ aliases for the old names. `--plan` output uses `match_by` in JSON and
18
+ "match by" in text. The stream format is unchanged.
19
+ - Query a plain planned `has_many` once per batch of sibling records, not
20
+ once per record. Siblings now come before their collections
21
+ in the stream. Any other planned collection is still queried per record.
22
+
23
+ ### Fixed
24
+
25
+ - Sync a primary key that is not auto-populated like any other field.
26
+ Deckard dropped every primary key, so loading such a row failed.
27
+
28
+ ## [0.3.1] - 2026-10-10
29
+
30
+ ### Fixed
31
+
32
+ - Reject unknown names in `omit_associations` with a `ConfigurationError`,
33
+ so a typo cannot silently leave an association included in a dump.
34
+
35
+ ## [0.3.0] - 2026-09-30
9
36
 
10
37
  ### Added
11
38
 
data/README.md CHANGED
@@ -2,8 +2,8 @@
2
2
 
3
3
  Deckard copies selected ActiveRecord records, and the records they need,
4
4
  from one Rails environment into another. Pick a record, and its plan says
5
- what comes with it. Rows land with new primary keys, foreign keys are
6
- rewritten to match, and the load commits only when the stream ends cleanly.
5
+ what comes with it. Deckard remaps foreign keys to the destination rows
6
+ and commits the load only when the stream ends cleanly.
7
7
  Its usual job is pulling one production record and its graph into a local
8
8
  development database to debug against real data.
9
9
 
@@ -34,7 +34,7 @@ class Order < ActiveRecord::Base
34
34
  end
35
35
 
36
36
  model "Customer" do
37
- natural_key :email
37
+ match_by :email
38
38
  omit_fields :password_digest
39
39
  end
40
40
  end
@@ -97,11 +97,11 @@ The `replicate` block offers six methods:
97
97
  ```ruby
98
98
  replicate do
99
99
  associations :line_items # opt this has_many into dumps
100
- natural_key :number # reuse an existing destination row
100
+ match_by :number # reuse an existing destination row
101
101
  omit_fields :internal_notes # keep a field out of the stream
102
102
  omit_associations :warehouse # do not traverse this association
103
103
 
104
- natural_keys "Product" => :sku, # how each reached class is matched
104
+ match_each_model_by "Product" => :sku, # how each reached class is matched
105
105
  "Warehouse" => [:region, :code]
106
106
 
107
107
  model "LineItem" do # the first four, for a class the dump reaches
@@ -110,15 +110,15 @@ replicate do
110
110
  end
111
111
  ```
112
112
 
113
- Most reached classes need nothing but a natural key, and `natural_keys` says
114
- that for all of them in one table. A `model` block is for the rest: a
115
- collection to bring along, a field to leave out. A plan gives each class one
116
- key, so naming it in both places is a plan problem, not an override.
113
+ `match_each_model_by` gives reached classes their match keys in one table. A
114
+ `model` block is for the rest: a collection to bring along, a field to leave
115
+ out. A plan gives each class one key, so naming it in both places is a plan
116
+ problem, not an override.
117
117
 
118
- Classes are named as strings, so the root never forces them to load first,
118
+ Classes are named as strings, so the root never autoloads them,
119
119
  and an STI subclass follows the entry for its nearest declared ancestor. Each
120
- record travels with the natural key its plan gave it, so the destination
121
- needs no plan of its own. That also means a natural key is repeated in every
120
+ record travels with the match key its plan gave it, so the destination
121
+ needs no plan of its own. That also means a match key is repeated in every
122
122
  plan whose dumps reach the class: a Customer line in Order's plan and one in
123
123
  Invoice's, each saying how a customer is matched.
124
124
 
@@ -143,7 +143,7 @@ LineItem
143
143
  associations adjustments
144
144
 
145
145
  Customer
146
- natural key email
146
+ match by email
147
147
  omit fields password_digest
148
148
  ```
149
149
 
@@ -181,6 +181,9 @@ terminal, a live object counter shows progress on both ends of the pipe.
181
181
  bypassing validations and callbacks: it copies data, it does not run your
182
182
  application. If the source dies mid-stream, nothing is committed.
183
183
 
184
+ Deckard syncs a primary key that is not auto-populated like any other
185
+ field. Give that model a match key if the same dump will be loaded twice.
186
+
184
187
  Loading refuses to run when the application environment is production. Pass
185
188
  `--force` to override.
186
189
 
@@ -196,7 +199,7 @@ plaintext and are re-encrypted with the destination's keys on load.
196
199
  ## Development
197
200
 
198
201
  `bundle exec rake` runs the unit/integration suite (specs against a local
199
- PostgreSQL 18, lint, and a style ratchet); `bin/rspec spec/deckard/stream_spec.rb`
202
+ PostgreSQL server, lint, and a style ratchet); `bin/rspec spec/deckard/stream_spec.rb`
200
203
  runs one file. The specs run against a small forum
201
204
  schema managed with ActiveRecord's own migration and schema tooling under
202
205
  `spec/db`; `bundle exec rake -T db` lists the database tasks, and
@@ -207,7 +210,12 @@ host-side checks and the full-stack test. See `docs/spec.md` for the v1.0
207
210
  specification and `docs/traversal.md` for how a dump walks the graph.
208
211
 
209
212
  Crow CI runs `bundle exec rake` on pull requests, default-branch pushes, and
210
- manual runs using Ruby 4.0.6 and an isolated PostgreSQL 18.6 service. The
211
- workflow is in `.crow/ruby.yaml`; it covers specs, Standard lint, and the
212
- style ratchet. The Docker end-to-end harness and performance suite remain
213
- separate local tasks.
213
+ manual runs using Ruby and an isolated PostgreSQL service. The workflow in
214
+ [`.crow/ruby.yaml`](.crow/ruby.yaml) defines the runtime versions and covers
215
+ specs, Standard lint, and the style ratchet. [`.crow/e2e.yaml`](.crow/e2e.yaml)
216
+ runs the end-to-end stream test. The performance suite remains a separate
217
+ local task.
218
+
219
+ See the [shared release guide](https://github.com/rsanheim/.github#readme) for
220
+ setup and operation, and the [Deckard quick reference](docs/releasing.md) for
221
+ the files to update in a release PR.
@@ -2,8 +2,8 @@
2
2
 
3
3
  module Deckard
4
4
  module ActiveRecord
5
- # Dumps one record by the root's plan (docs/traversal.md), and the records
6
- # it reaches by the same plan.
5
+ # Dumps records by the root's plan (docs/traversal.md), and the records
6
+ # they reach by the same plan.
7
7
  class Dump
8
8
  # Stored generated columns (GENERATED ALWAYS AS ... STORED) are never
9
9
  # dumped: the destination database computes them, and PostgreSQL
@@ -19,6 +19,30 @@ module Deckard
19
19
  record.id
20
20
  end
21
21
 
22
+ # Dumps sibling records the plan owns, then each planned collection of
23
+ # the ones this visit walked.
24
+ def self.dump_owned(records, dumper, plan)
25
+ walked = records.select { |record| new(record, dumper, plan, owned: true).call }
26
+ walked.group_by(&:class).each do |model, owners|
27
+ entry = plan.for(model)
28
+ (entry.extra_associations - entry.omitted_associations).each do |name|
29
+ reflection = model.reflect_on_association(name)
30
+ next unless reflection.collection?
31
+
32
+ # Deckard loads a plain has_many for all the siblings in one query;
33
+ # anything else is queried per record.
34
+ plain = reflection.macro == :has_many && reflection.scope.nil? && reflection.klass.default_scopes.empty? &&
35
+ (reflection.options.keys - %i[class_name foreign_key inverse_of dependent]).empty?
36
+ relations = if plain
37
+ [reflection.klass.where(reflection.foreign_key => owners.map(&:id))]
38
+ else
39
+ owners.map { |owner| owner.public_send(name) }
40
+ end
41
+ relations.each { |relation| relation.find_in_batches { |batch| dump_owned(batch, dumper, plan) } }
42
+ end
43
+ end
44
+ end
45
+
22
46
  def initialize(record, dumper, plan, owned:)
23
47
  @record = record
24
48
  @model = record.class
@@ -29,16 +53,25 @@ module Deckard
29
53
  @entry = plan.for(@model)
30
54
  end
31
55
 
56
+ # Dumps the record and, when owned, its singular owned records. Planned
57
+ # collections are left to dump_owned. True when this visit walked it.
32
58
  def call
59
+ walked = false
33
60
  @dumper.visit(@model.name, @source_id, walk: @owned) do
34
61
  attributes = @record.attributes.except(*GENERATED_COLUMNS[@model], *@entry.omitted_fields)
35
62
  dump_belongs_to(attributes)
36
- @dumper.write(@model.name, @source_id, attributes, @entry.natural_key_attributes)
63
+ @dumper.write(@model.name, @source_id, attributes, @entry.match_key)
37
64
  next unless @owned
38
65
 
66
+ walked = true
39
67
  owned = @model.reflect_on_all_associations(:has_one).map(&:name) | @entry.extra_associations
40
- (owned - @entry.omitted_associations).each { |name| dump_owned(@record.public_send(name)) }
68
+ (owned - @entry.omitted_associations).each do |name|
69
+ next if @model.reflect_on_association(name).collection?
70
+
71
+ self.class.dump_owned(Array(@record.public_send(name)), @dumper, @plan)
72
+ end
41
73
  end
74
+ walked
42
75
  end
43
76
 
44
77
  private
@@ -61,8 +94,8 @@ module Deckard
61
94
  end
62
95
  next if @entry.omitted_fields.include?(foreign_key)
63
96
  # A belongs_to whose primary_key option targets a non-primary-key
64
- # column (belongs_to :account, primary_key: :login) carries a natural
65
- # value, not a source ID. It needs no remapping and must not be
97
+ # column (belongs_to :account, primary_key: :login) holds that
98
+ # column's value, not a source ID. It needs no remapping and must not be
66
99
  # replaced with a destination primary key.
67
100
  next unless reflection.association_primary_key(referenced.class) == referenced.class.primary_key
68
101
 
@@ -84,11 +117,6 @@ module Deckard
84
117
  attributes[foreign_key] = [:id, reflection.klass.name, value] unless @entry.omitted_fields.include?(foreign_key)
85
118
  true
86
119
  end
87
-
88
- def dump_owned(associated)
89
- records = associated.respond_to?(:find_each) ? associated.find_each : Array(associated)
90
- records.each { |record| self.class.new(record, @dumper, @plan, owned: true).call }
91
- end
92
120
  end
93
121
  end
94
122
  end
@@ -4,12 +4,12 @@ module Deckard
4
4
  module ActiveRecord
5
5
  # Loads one replicant tuple into a model's table.
6
6
  class Load
7
- def initialize(model, type, source_id, attributes, natural_key)
7
+ def initialize(model, type, source_id, attributes, match_key)
8
8
  @model = model
9
9
  @type = type
10
10
  @source_id = source_id
11
11
  @attributes = attributes
12
- @natural_key = natural_key
12
+ @match_key = match_key
13
13
  end
14
14
 
15
15
  def call
@@ -17,15 +17,18 @@ module Deckard
17
17
  raise LoadError, "#{@model.name} has a composite primary key, which deckard does not support"
18
18
  end
19
19
 
20
- @natural_key.empty? ? insert : load_by_natural_key
20
+ @match_key.empty? ? insert : load_by_match_key
21
21
  end
22
22
 
23
23
  private
24
24
 
25
25
  # `because` says what led to the insert when it was not the only path,
26
- # so a failure names the natural key that matched nothing.
26
+ # so a failure names the match key that matched nothing.
27
27
  def insert(because = "")
28
- result = @model.insert_all!([@attributes.except(@model.primary_key)], returning: [@model.primary_key])
28
+ # An auto-populated primary key is left to the destination database.
29
+ auto_populated = @model.columns_hash[@model.primary_key].auto_populated?
30
+ attributes = auto_populated ? @attributes.except(@model.primary_key) : @attributes
31
+ result = @model.insert_all!([attributes], returning: [@model.primary_key])
29
32
  destination_id = result.rows.first.first
30
33
  [destination_id, @model.find(destination_id)]
31
34
  rescue ::ActiveRecord::ActiveRecordError => e
@@ -33,25 +36,25 @@ module Deckard
33
36
  "#{@type} source_id=#{@source_id} #{because}could not be inserted: #{Deckard.error_detail(e)}"
34
37
  end
35
38
 
36
- def load_by_natural_key
37
- lookup = @attributes.slice(*@natural_key)
39
+ def load_by_match_key
40
+ lookup = @attributes.slice(*@match_key)
38
41
  matches = @model.where(lookup).limit(2).to_a
39
42
 
40
43
  case matches.size
41
44
  when 0
42
- insert("matched no destination row by natural key (#{lookup.keys.join(", ")}) and ")
45
+ insert("matched no destination row by match key (#{lookup.keys.join(", ")}) and ")
43
46
  when 1
44
47
  record = matches.first
45
48
  begin
46
49
  record.update_columns(@attributes.except(@model.primary_key))
47
50
  rescue ::ActiveRecord::ActiveRecordError => e
48
51
  raise InsertError,
49
- "#{@type} source_id=#{@source_id} could not be updated via natural key: #{Deckard.error_detail(e)}"
52
+ "#{@type} source_id=#{@source_id} could not be updated via match key: #{Deckard.error_detail(e)}"
50
53
  end
51
54
  [record.id, record]
52
55
  else
53
56
  raise LoadError,
54
- "#{@type} source_id=#{@source_id}: natural key (#{lookup.keys.join(", ")}) matches more than one destination row"
57
+ "#{@type} source_id=#{@source_id}: match key (#{lookup.keys.join(", ")}) matches more than one destination row"
55
58
  end
56
59
  end
57
60
  end
@@ -16,7 +16,7 @@ module Deckard
16
16
  # Records reached from this one are dumped by its plan directly, so an
17
17
  # override of this method applies to roots only.
18
18
  def dump_replicant(dumper)
19
- Dump.new(self, dumper, ModelConfig.plan_for(self.class), owned: true).call
19
+ Dump.dump_owned([self], dumper, ModelConfig.plan_for(self.class))
20
20
  end
21
21
 
22
22
  module ClassMethods
@@ -25,11 +25,11 @@ module Deckard
25
25
  ModelConfig.declare(name, &block)
26
26
  end
27
27
 
28
- # Load one streamed replicant: reuse an existing row when its natural
29
- # key matches, otherwise insert a new row with a destination-generated
30
- # primary key. Both paths bypass validations and callbacks.
31
- def load_replicant(type, source_id, attributes, natural_key)
32
- Load.new(self, type, source_id, attributes, natural_key).call
28
+ # Load one streamed replicant: reuse an existing row when its match
29
+ # key matches, otherwise insert a new row. Both paths bypass validations
30
+ # and callbacks.
31
+ def load_replicant(type, source_id, attributes, match_key)
32
+ Load.new(self, type, source_id, attributes, match_key).call
33
33
  end
34
34
  end
35
35
  end
@@ -57,13 +57,13 @@ module Deckard
57
57
  end
58
58
 
59
59
  # Called by dump_replicant implementations to emit one replicant tuple.
60
- # natural_key names the attributes by which the destination matches an
60
+ # match_key names the attributes by which the destination matches an
61
61
  # existing record to reuse; empty means always insert.
62
- def write(type, id, attributes, natural_key = [])
62
+ def write(type, id, attributes, match_key = [])
63
63
  type = type.to_s
64
64
  return unless @dumped.add?([type, id])
65
65
 
66
- write_frame([type, id, attributes, natural_key])
66
+ write_frame([type, id, attributes, match_key])
67
67
  @counts[type] += 1
68
68
  @after_write&.call(@counts)
69
69
  end
@@ -3,7 +3,7 @@
3
3
  module Deckard
4
4
  # Reads a Deckard stream incrementally from an input IO, resolving
5
5
  # [:id, type, source_id] reference tuples through the source-to-destination
6
- # ID map and handing each replicant, with the natural key it travels
6
+ # ID map and handing each replicant, with the match key it travels
7
7
  # with, to its class's load_replicant.
8
8
  class Loader
9
9
  attr_reader :counts
@@ -58,12 +58,12 @@ module Deckard
58
58
 
59
59
  def load_replicant(frame)
60
60
  unless frame.is_a?(Array) && frame.size == 4 && frame[0].is_a?(String) && frame[2].is_a?(Hash) && frame[3].is_a?(Array)
61
- raise InvalidStream, "malformed stream frame (expected [type, id, attributes, natural_key] tuple)"
61
+ raise InvalidStream, "malformed stream frame (expected [type, id, attributes, match_key] tuple)"
62
62
  end
63
63
 
64
- type, source_id, attributes, natural_key = frame
64
+ type, source_id, attributes, match_key = frame
65
65
  resolved = resolve_references(type, source_id, attributes)
66
- destination_id, _object = replicant_class(type).load_replicant(type, source_id, resolved, natural_key)
66
+ destination_id, _object = replicant_class(type).load_replicant(type, source_id, resolved, match_key)
67
67
  @id_map[[type, source_id]] = destination_id
68
68
  @counts[type] += 1
69
69
  @after_load&.call(@counts)
@@ -5,11 +5,11 @@ require "active_support/core_ext/string/inflections"
5
5
  module Deckard
6
6
  # A replication plan: a root model's `replicate do ... end` block. The plan
7
7
  # is an entry for the root plus an entry for each class its dumps reach,
8
- # declared through `natural_keys` or `model`, all in one table shared by
9
- # the plan's entries. A class finds its entry through its ancestors, so an
10
- # STI subclass follows its nearest declared ancestor. See docs/traversal.md.
8
+ # declared through `match_each_model_by` or `model`, all in one table
9
+ # shared by the plan's entries. A class finds its entry through its
10
+ # ancestors, so an STI subclass follows its nearest declared ancestor. See docs/traversal.md.
11
11
  class ModelConfig
12
- attr_reader :name, :extra_associations, :natural_key_attributes, :omitted_fields, :omitted_associations
12
+ attr_reader :name, :extra_associations, :match_key, :omitted_fields, :omitted_associations
13
13
 
14
14
  @plans = {}
15
15
 
@@ -35,7 +35,7 @@ module Deckard
35
35
  @entries[name] = self if name
36
36
  @resolved = {}
37
37
  @extra_associations = []
38
- @natural_key_attributes = []
38
+ @match_key = []
39
39
  @omitted_fields = []
40
40
  @omitted_associations = []
41
41
  end
@@ -48,14 +48,14 @@ module Deckard
48
48
 
49
49
  # DSL: attributes identifying an existing destination record to reuse.
50
50
  # Declaring it twice in one plan is a problem, not an override.
51
- def natural_key(*attributes)
52
- @redeclared_natural_key = !@natural_key_attributes.empty?
53
- @natural_key_attributes = attributes.map(&:to_s)
51
+ def match_by(*attributes)
52
+ @redeclared_match_key = !@match_key.empty?
53
+ @match_key = attributes.map(&:to_s)
54
54
  end
55
55
 
56
- # DSL: class name to natural key, for reached classes needing nothing else.
57
- def natural_keys(keys)
58
- keys.each { |name, attributes| model(name) { natural_key(*attributes) } }
56
+ # DSL: class name to match key, for reached classes needing nothing else.
57
+ def match_each_model_by(keys)
58
+ keys.each { |name, attributes| model(name) { match_by(*attributes) } }
59
59
  end
60
60
 
61
61
  # DSL: fields to exclude from dumped attributes. Additive across calls.
@@ -105,15 +105,18 @@ module Deckard
105
105
  "which deckard does not support; replicate :#{reflection.through_reflection.name} instead"
106
106
  end
107
107
  end
108
- attributes = @natural_key_attributes + @omitted_fields
108
+ problems += @omitted_associations.filter_map do |name|
109
+ "#{model} has no #{name.inspect} association" unless model.reflect_on_association(name)
110
+ end
111
+ attributes = @match_key + @omitted_fields
109
112
  unless attributes.empty? || model.abstract_class?
110
113
  missing = attributes - model.attribute_names
111
114
  problems += missing.map { |attribute| "#{model} has no #{attribute.inspect} attribute" }
112
115
  end
113
- problems += (@natural_key_attributes & @omitted_fields).map do |attribute|
114
- "#{model} names #{attribute.inspect} as both a natural key attribute and an omitted field"
116
+ problems += (@match_key & @omitted_fields).map do |attribute|
117
+ "#{model} names #{attribute.inspect} as both a match key attribute and an omitted field"
115
118
  end
116
- problems << "#{model} is given a natural key more than once in the same plan" if @redeclared_natural_key
119
+ problems << "#{model} is given a match key more than once in the same plan" if @redeclared_match_key
117
120
  problems
118
121
  end
119
122
 
@@ -123,7 +126,7 @@ module Deckard
123
126
  {
124
127
  "model" => entry.name,
125
128
  "associations" => entry.extra_associations.map(&:to_s),
126
- "natural_key" => entry.natural_key_attributes,
129
+ "match_by" => entry.match_key,
127
130
  "omit_fields" => entry.omitted_fields,
128
131
  "omit_associations" => entry.omitted_associations.map(&:to_s)
129
132
  }
@@ -10,7 +10,7 @@ module Deckard
10
10
 
11
11
  LINES = {
12
12
  "associations" => "associations",
13
- "natural_key" => "natural key",
13
+ "match_by" => "match by",
14
14
  "omit_fields" => "omit fields",
15
15
  "omit_associations" => "omit associations"
16
16
  }.freeze
@@ -1,5 +1,5 @@
1
1
  # frozen_string_literal: true
2
2
 
3
3
  module Deckard
4
- VERSION = "0.3.0"
4
+ VERSION = "0.4.1"
5
5
  end
metadata CHANGED
@@ -1,7 +1,7 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: deckard
3
3
  version: !ruby/object:Gem::Version
4
- version: 0.3.0
4
+ version: 0.4.1
5
5
  platform: ruby
6
6
  authors:
7
7
  - Rob Sanheim
@@ -44,8 +44,7 @@ dependencies:
44
44
  - !ruby/object:Gem::Version
45
45
  version: '3.2'
46
46
  description: Copy selected ActiveRecord records and their associations between PostgreSQL
47
- databases, with destination-generated primary keys, remapped foreign keys, and transactional
48
- loads.
47
+ databases, with remapped foreign keys and transactional loads.
49
48
  email:
50
49
  - rsanheim@gmail.com
51
50
  executables:
@@ -90,7 +89,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
90
89
  - !ruby/object:Gem::Version
91
90
  version: '0'
92
91
  requirements: []
93
- rubygems_version: 4.0.21
92
+ rubygems_version: 4.0.20
94
93
  specification_version: 4
95
94
  summary: Stream ActiveRecord objects between Rails environments
96
95
  test_files: []