factbase 0.19.8 → 0.19.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 03d4be418b18984640fa333733a6b688aa2879b2e7def6d0ad901bce47490cd2
-  data.tar.gz: c6f24e2c444eb03eb739489d4cd09f927ece90a8c44f480c3b04fef1876b574d
+  metadata.gz: 4b573b12e8a04a0b69a2e4c33c6ccd17c21c1c14a52a644c87564738388de546
+  data.tar.gz: b4c888226e04aee8054ff26d056cc04f34fd3a1fb29455ff6c77ce4dd2103d45
 SHA512:
-  metadata.gz: f73778d27a87c5a436769565032c5b064b24354ce62fad61822fd29941664e96c6d70f8481bf146bc72ef32669fc4969bae050eb70ef2f5c1fa0e19ee9395000
-  data.tar.gz: 299d0afc68fa48166fd4e01366e82ce1d22f8aa8c2f1dc2cfc9bd397763080f8db1762b20caa7738bafd0a5d606e9fa1725c13f400e13b40193ff13528b05b59
+  metadata.gz: a96ff08cc40deb0d17e25cc7c30f8e67d5c09b0179e623d7e941ade194ac7c52c2135730fb4a813faf7370097a1d151db74e1559de81a619bf572becb1eec408
+  data.tar.gz: 1f57d718a82c2c8d107c033c3d8b59440170822f5810fadfc3b99837cf4072493e96ec8a39bff41c73a913cc3dba75837a9f4ecc0804976da37ebe02c09b1777

data/Gemfile.lock

@@ -16,7 +16,7 @@ PATH
 GEM
   remote: https://rubygems.org/
   specs:
-    ansi (1.5.0)
+    ansi (1.6.0)
     ast (2.4.3)
     backtrace (0.4.1)
     benchmark (0.5.0)
@@ -38,13 +38,13 @@ GEM
     loog (0.8.0)
       ellipsized
       logger (~> 1.0)
-    minitest (6.0.2)
+    minitest (6.0.5)
       drb (~> 2.0)
       prism (~> 1.5)
-    minitest-reporters (1.7.1)
+    minitest-reporters (1.8.0)
       ansi
       builder
-      minitest (>= 5.0)
+      minitest (>= 5.0, < 7)
       ruby-progressbar
     nokogiri (1.19.0-arm64-darwin)
       racc (~> 1.4)
@@ -64,14 +64,14 @@ GEM
     psych (5.3.1)
       date
       stringio
-    qbash (0.8.0)
+    qbash (0.8.3)
       backtrace (> 0)
       elapsed (> 0)
       loog (> 0)
       tago (> 0)
     racc (1.8.1)
     rainbow (3.1.1)
-    rake (13.3.1)
+    rake (13.4.2)
     rdoc (7.1.0)
       erb
       psych (>= 4.0.0)

data/README.md

@@ -95,6 +95,30 @@ fb.query('(eq foo 43)').each do |f|
 end
 ```
 
+Deleting while iterating is unsafe and may cause elements to be skipped:
+
+```ruby
+fb = Factbase.new
+fb.insert.id = 1
+fb.insert.id = 2
+fb.query('(always)').each do |f|
+  fb.query("(eq id #{f.id})").delete!
+end
+assert(1 == fb.size)
+```
+
+To safely delete, use a snapshot:
+
+```ruby
+fb = Factbase.new
+fb.insert.id = 1
+fb.insert.id = 2
+fb.query('(always)').to_a.each do |f|
+  fb.query("(eq id #{f.id})").delete!
+end
+assert(0 == fb.size)
+```
+
 ## Terms
 
 There are some boolean terms available in a query
@@ -225,91 +249,25 @@ This is the result of the benchmark:
 
 <!-- benchmark_begin -->
 ```text
-                                                                       
-query all facts from an empty factbase                             0.00
-insert 20000 facts                                                 0.63
-export 20000 facts                                                 0.02
-import 411003 bytes (20000 facts)                                  0.01
-insert 10 facts                                                    0.03
-query 10 times w/txn                                               2.03
-query 10 times w/o txn                                             0.13
-modify 10 attrs w/txn                                              1.51
-delete 10 facts w/txn                                              2.97
-build index on 5000 facts                                          0.03
-export 5000 facts with index                                       0.04
-import 5000 facts with persisted index                             0.03
-query 5000 facts using persisted index                             0.07
-export 5000 facts without index                                    0.01
-import 5000 facts without index                                    0.01
-query 5000 facts building index on-the-fly                         0.07
-query 15k facts  sel: 20%  card: 10  absent plain                  0.61
-query 15k facts  sel: 20%  card: 10  absent indexed(cold)          0.16
-query 15k facts  sel: 20%  card: 10  absent indexed(warm)          0.13
-query 15k facts  sel: 20%  card: 10  exists plain                  0.56
-query 15k facts  sel: 20%  card: 10  exists indexed(cold)          0.14
-query 15k facts  sel: 20%  card: 10  exists indexed(warm)          0.13
-query 15k facts  sel: 20%  card: 10  eq plain                      0.83
-query 15k facts  sel: 20%  card: 10  eq indexed(cold)              0.22
-query 15k facts  sel: 20%  card: 10  eq indexed(warm)              0.24
-query 15k facts  sel: 20%  card: 10  not plain                     1.11
-query 15k facts  sel: 20%  card: 10  not indexed(cold)             0.50
-query 15k facts  sel: 20%  card: 10  not indexed(warm)             0.43
-query 15k facts  sel: 20%  card: 10  gt plain                      0.84
-query 15k facts  sel: 20%  card: 10  gt indexed(cold)              0.26
-query 15k facts  sel: 20%  card: 10  gt indexed(warm)              0.19
-query 15k facts  sel: 20%  card: 10  lt plain                      0.84
-query 15k facts  sel: 20%  card: 10  lt indexed(cold)              0.27
-query 15k facts  sel: 20%  card: 10  lt indexed(warm)              0.20
-query 15k facts  sel: 20%  card: 10  and eq plain                  1.39
-query 15k facts  sel: 20%  card: 10  and eq indexed(cold)          0.85
-query 15k facts  sel: 20%  card: 10  and eq indexed(warm)          0.46
-query 15k facts  sel: 20%  card: 10  and complex plain             1.32
-query 15k facts  sel: 20%  card: 10  and complex indexed(cold)     0.48
-query 15k facts  sel: 20%  card: 10  and complex indexed(warm)     0.43
-query 15k facts  sel: 20%  card: 10  one plain                     0.72
-query 15k facts  sel: 20%  card: 10  one indexed(cold)             0.19
-query 15k facts  sel: 20%  card: 10  one indexed(warm)             0.15
-query 15k facts  sel: 20%  card: 10  or plain                      2.02
-query 15k facts  sel: 20%  card: 10  or indexed(cold)              0.42
-query 15k facts  sel: 20%  card: 10  or indexed(warm)              0.36
-query 15k facts  sel: 20%  card: 10  unique plain                  1.86
-query 15k facts  sel: 20%  card: 10  unique indexed(cold)          0.61
-query 15k facts  sel: 20%  card: 10  unique indexed(warm)          0.37
-(and (eq what 'issue-was-closed') (exists... -> 200                1.04
-(and (eq what 'issue-was-closed') (exists... -> 200/txn            1.23
-(and (eq what 'issue-was-closed') (exists... -> zero               1.06
-(and (eq what 'issue-was-closed') (exists... -> zero/txn           1.23
-transaction rollback on factbase with 100000 facts                 0.26
-(gt time '2024-03-23T03:21:43Z')                                   0.22
-(gt cost 50)                                                       0.10
-(eq title 'Object Thinking 5000')                                  0.03
-(and (eq foo 42.998) (or (gt bar 200) (absent z...                 0.02
-(and (exists foo) (not (exists blue)))                             1.12
-(eq id (agg (always) (max id)))                                    2.76
-(join "c<=cost,b<=bar" (eq id (agg (always) (ma...                 4.58
-(and (eq what "foo") (join "w<=what" (and (eq i...                 7.26
-delete!                                                            0.42
-(and (eq issue *) (eq repository *) (eq what '*') (eq where '*'))  0.38
-Taped.append() x50000                                              0.02
-Taped.each() x125                                                  1.08
-Taped.delete_if() x375                                             0.86
-50000 facts: plain read (no txn)                                   4.06
-50000 facts: read-only txn (no copy)                               4.85
-50000 facts: plain insert (no txn)                                 0.00
-50000 facts: insert in txn (copy triggered)                        3.31
-50000 facts: plain modify (no txn)                                28.21
-50000 facts: modify in txn (copy triggered)                       35.14
-100000 facts: plain read (no txn)                                  8.06
-100000 facts: read-only txn (no copy)                              9.94
-100000 facts: plain insert (no txn)                                0.00
-100000 facts: insert in txn (copy triggered)                       6.68
-100000 facts: plain modify (no txn)                               56.21
-100000 facts: modify in txn (copy triggered)                      70.09
+                                                                   user
+void scan                                                      0.001102
+20k facts: export: 2991KB                                      0.854774
+20k facts: import: 2991KB                                      1.035671
+50k facts: read                                                0.000138
+50k facts: read in txn                                         0.002754
+50k facts: insert                                              0.000090
+50k facts: insert in txn                                       0.000243
+50k facts: modify                                              1.085214
+50k facts: modify in txn                                       2.480409
+12k facts: large query: match 3k                              14.187596
+12k facts: large query: match 3k in txn                       19.396334
+12k facts: large query: match zero                            15.139695
+12k facts: large query: match zero in txn                     21.046074
 ```
 
 The results were calculated in [this GHA job][benchmark-gha]
-on 2026-02-17 at 18:48,
+on 2026-04-16 at 17:08,
 on Linux with 4 CPUs.
 <!-- benchmark_end -->
 
-[benchmark-gha]: https://github.com/yegor256/factbase/actions/runs/22111136309
+[benchmark-gha]: https://github.com/yegor256/factbase/actions/runs/24523476155

data/Rakefile

@@ -20,6 +20,12 @@ end
 
 task default: %i[clean test picks rubocop yard]
 
+def tail(args = ARGV)
+  i = args.index('--')
+  return '' if i.nil? || args[i + 1].nil?
+  args[i..].join(' ')
+end
+
 require 'rake/testtask'
 desc 'Run all unit tests'
 Rake::TestTask.new(:test) do |test|
@@ -28,6 +34,7 @@ Rake::TestTask.new(:test) do |test|
   test.pattern = 'test/**/test_*.rb'
   test.warning = true
   test.verbose = false
+  test.options = tail
 end
 
 desc 'Run them via Ruby, one by one'
@@ -54,27 +61,30 @@ RuboCop::RakeTask.new(:rubocop) do |task|
 end
 
 desc 'Benchmark them all'
-task :benchmark, [:name] do |_t, args|
-  bname = args[:name] || 'all'
+task :benchmark, [:name, :cycles] do |_t, args|
+  bname = args[:name] || 'essential'
+  cycles = (args[:cycles] || 5).to_i
   require_relative 'lib/factbase'
   require_relative 'lib/factbase/cached/cached_factbase'
   require_relative 'lib/factbase/indexed/indexed_factbase'
   require_relative 'lib/factbase/sync/sync_factbase'
   require 'benchmark'
   Benchmark.bm(60) do |b|
-    fb = Factbase.new
-    fb = Factbase::CachedFactbase.new(fb)
-    fb = Factbase::IndexedFactbase.new(fb)
-    fb = Factbase::SyncFactbase.new(fb)
-    if bname == 'all'
-      Dir['benchmark/bench_*.rb'].each do |f|
-        require_relative f
-        Kernel.send(File.basename(f).gsub(/\.rb$/, '').to_sym, b, fb)
+    files =
+      case bname
+      when 'all'
+        Dir['benchmark/bench_*.rb']
+      when 'essential'
+        %w[empty serialization txns large_query].map { |f| "benchmark/bench_#{f}.rb" }
+      else
+        ["benchmark/#{bname}.rb"]
       end
-    else
-      f = "benchmark/#{bname}.rb"
+    files.each do |f|
       require_relative f
-      Kernel.send(File.basename(f).gsub(/\.rb$/, '').to_sym, b, fb)
+      fb = Factbase.new
+      fb = Factbase::IndexedFactbase.new(fb)
+      fb = Factbase::CachedFactbase.new(fb)
+      Kernel.send(File.basename(f).gsub(/\.rb$/, '').to_sym, b, fb, cycles)
     end
   end
 end

data/lib/factbase/indexed/indexed_and.rb

@@ -51,8 +51,9 @@ class Factbase::IndexedAnd
         if r.nil?
           r = n
         elsif n.size < r.size * 8 # to skip some obvious matchings
-          ids = n.to_set(&:object_id)
-          r = r.select { |f| ids.include?(f.object_id) }
+          small, large = n.size < r.size ? [n.to_a, r.to_a] : [r.to_a, n.to_a]
+          ids = Set.new(small.map(&:object_id))
+          r = large.select { |f| ids.include?(f.object_id) }
         end
         break if r.size < maps.size / 32 # it's already small enough
         break if r.size < 128 # it's obviously already small enough

data/lib/factbase/indexed/indexed_factbase.rb

@@ -61,11 +61,12 @@ class Factbase::IndexedFactbase
   # Run an ACID transaction.
   # @return [Factbase::Churn] How many facts have been changed (zero if rolled back)
   def txn
+    inner_idx = {}
     result =
       @origin.txn do |fbt|
-        yield Factbase::IndexedFactbase.new(fbt, @idx, @fresh)
+        yield Factbase::IndexedFactbase.new(fbt, inner_idx, @fresh)
       end
-    @idx.clear
+    @idx.clear if result.deleted.positive? || result.added.positive?
     @fresh.clear
     result
   end

data/lib/factbase/indexed/indexed_unique.rb

@@ -4,23 +4,6 @@
 # SPDX-License-Identifier: MIT
 
 # Indexed term 'unique'.
-# The @idx[ikey] structure:
-# {
-#   count: Integer (number of facts already processed),
-#   buckets: {
-#     key => {
-#       facts: Array (unique facts found),
-#       seen: Set (composite values already indexed to skip duplicates)
-#     }
-#   }
-# }
-# Example 1: (unique "fruit")
-#   - Apple, Apple, Banana
-#   - count: 3, facts: [Apple, Banana], seen: { [Apple], [Banana] }
-#
-# Example 2: (unique "fruit" "color")
-#   - [Apple, Red], [Apple, Green], [Apple, Red]
-#   - count: 3, facts: [[Apple, Red], [Apple, Green]], seen: { [Apple, Red], [Apple, Green] }
 class Factbase::IndexedUnique
   def initialize(term, idx)
     @term = term
@@ -28,25 +11,6 @@ class Factbase::IndexedUnique
   end
 
   def predict(maps, _fb, _params)
-    operands = @term.operands.map(&:to_s)
-    bucket_key = operands.join('|')
-    idx_key = [maps.object_id, @term.op.to_s, bucket_key]
-    entry = (@idx[idx_key] ||= { buckets: {}, count: 0 })
-    feed(maps.to_a, entry, operands, bucket_key)
-    matches = entry[:buckets][bucket_key][:facts]
-    maps.respond_to?(:repack) ? maps.repack(matches) : matches
-  end
-
-  private
-
-  def feed(facts, entry, operands, bucket_key)
-    entry[:buckets][bucket_key] ||= { facts: [], seen: Set.new }
-    bucket = entry[:buckets][bucket_key]
-    (facts[entry[:count]..] || []).each do |fact|
-      composite_val = operands.map { |o| fact[o] }
-      next if composite_val.any?(&:nil?)
-      bucket[:facts] << fact if bucket[:seen].add?(composite_val)
-    end
-    entry[:count] = facts.size
+    maps
   end
 end

data/lib/factbase/lazy_taped.rb

@@ -72,17 +72,18 @@ class Factbase::LazyTaped
 
   def each
     return to_enum(__method__) unless block_given?
-    st_size = @staged.size
-    orig_size = @origin.size
-    unless copied?
-      orig_size.times do |i|
-        m = @origin[i]
-        yield _tape(m) unless m.nil?
+    yielded_size = 0
+    is_copied = copied?
+    unless is_copied
+      @origin.each do |m|
+        yield _tape(m)
+        yielded_size += 1
       end
     end
-    st_size.times do |i|
-      m = @staged[i]
-      yield _tape(m) unless m.nil?
+    staged = is_copied == copied? ? @staged : @staged[yielded_size..]
+    staged&.each do |f|
+      next if f.nil?
+      yield _tape(f)
     end
   end
 
@@ -119,11 +120,9 @@ class Factbase::LazyTaped
 
   def ensure_copied!
     return if copied?
-    @origin.each do |o|
-      c = o.transform_values(&:dup)
-      _track(c, o)
-      @staged << c
-    end
+    @staged = @origin.map do |o|
+      o.transform_values(&:dup).tap { |c| _track(c, o) }
+    end.concat(@staged)
     @copied = true
   end
 

data/lib/factbase/version.rb

@@ -9,5 +9,5 @@
 # License:: MIT
 class Factbase
   # Current version of the gem (changed by .rultor.yml on every release)
-  VERSION = '0.19.8' unless const_defined?(:VERSION)
+  VERSION = '0.19.10' unless const_defined?(:VERSION)
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: factbase
 version: !ruby/object:Gem::Version
-  version: 0.19.8
+  version: 0.19.10
 platform: ruby
 authors:
 - Yegor Bugayenko