bud 0.9.7 → 0.9.8

data/lib/bud/collections.rb

@@ -1,4 +1,3 @@
-$struct_classes = {}
 module Bud
   ########
   #--
@@ -17,7 +16,7 @@ module Bud
     attr_accessor :bud_instance  # :nodoc: all
     attr_reader :tabname, :cols, :key_cols # :nodoc: all
     attr_reader :struct
-    attr_reader :storage, :delta, :new_delta, :pending, :tick_delta # :nodoc: all
+    attr_reader :new_delta, :pending # :nodoc: all
     attr_reader :wired_by, :scanner_cnt
     attr_accessor :invalidated, :rescan
     attr_accessor :is_source
@@ -63,7 +62,7 @@ module Bud
       if @cols.empty?
         @cols = nil
       else
-        @struct = ($struct_classes[@cols] ||= Bud::TupleStruct.new(*@cols))
+        @struct = Bud::TupleStruct.new_struct(@cols)
         @structlen = @struct.members.length
       end
       setup_accessors
@@ -250,7 +249,7 @@ module Bud
     def sort(&blk)
       if @bud_instance.wiring?
         pusher = self.pro
-        pusher.sort("sort#{object_id}", @bud_instance, @cols, &blk)
+        pusher.sort("sort#{object_id}".to_sym, @bud_instance, @cols, &blk)
       else
         @storage.values.sort(&blk)
       end
@@ -275,7 +274,12 @@ module Bud
 
     public
     def each_raw(&block)
-      @storage.each_value(&block)
+      each_from([@storage], &block)
+    end
+
+    public
+    def each_delta(&block)
+      each_from([@delta], &block)
     end
 
     public
@@ -301,37 +305,26 @@ module Bud
     public
     def tick_metrics
       strat_num = bud_instance.this_stratum
-      rule_num = bud_instance.this_rule
-      addr = nil
       addr = bud_instance.ip_port unless bud_instance.port.nil?
+      key = { :addr=>addr, :tabname=>qualified_tabname,
+              :strat_num=>strat_num}
+
       bud_instance.metrics[:collections] ||= {}
-      bud_instance.metrics[:collections][{:addr=>addr, :tabname=>qualified_tabname, :strat_num=>strat_num, :rule_num=>rule_num}] ||= 0
-      bud_instance.metrics[:collections][{:addr=>addr, :tabname=>qualified_tabname, :strat_num=>strat_num, :rule_num=>rule_num}] += 1
+      bud_instance.metrics[:collections][key] ||= 0
+      bud_instance.metrics[:collections][key] += 1
     end
 
     private
     def each_from(bufs, &block) # :nodoc: all
+      do_metrics = bud_instance.options[:metrics]
       bufs.each do |b|
         b.each_value do |v|
-          tick_metrics if bud_instance and bud_instance.options[:metrics]
+          tick_metrics if do_metrics
           yield v
         end
       end
     end
 
-    public
-    def each_from_sym(buf_syms, &block) # :nodoc: all
-      bufs = buf_syms.map do |s|
-        case s
-        when :storage then @storage
-        when :delta then @delta
-        when :new_delta then @new_delta
-        else raise Bud::Error, "bad symbol passed into each_from_sym"
-        end
-      end
-      each_from(bufs, &block)
-    end
-
     private
     def init_storage
       @storage = {}
@@ -374,7 +367,7 @@ module Bud
     # checks for +item+ in the collection
     public
     def include?(item)
-      return true if key_cols.nil? or (key_cols.empty? and length > 0)
+      return true if key_cols.nil?
       return false if item.nil?
       key = get_key_vals(item)
       return (item == self[key])
@@ -650,6 +643,11 @@ module Bud
       end
     end
 
+    superator "<~" do |o|
+      # Overridden when <~ is defined (i.e., channels and terminals)
+      raise Bud::CompileError, "#{tabname} cannot appear on the lhs of a <~ operator"
+    end
+
     def tick
       raise Bud::Error, "tick must be overriden in #{self.class}"
     end
@@ -1111,6 +1109,11 @@ module Bud
       return true
     end
 
+    public
+    def bootstrap
+      # override BudCollection; pending should not be moved into delta.
+    end
+
     public
     def flush #:nodoc: all
       out_io = get_out_io
@@ -1257,7 +1260,7 @@ module Bud
 
     def invalidated=(val)
       # Might be reset to false at end-of-tick, but shouldn't be set to true
-      raise Bud::Error, "cannot not set invalidate on table '#{@tabname}'" if val
+      raise Bud::Error, "cannot set invalidate on table '#{@tabname}'" if val
       super
     end
 
@@ -1383,7 +1386,7 @@ module Bud
   end
 
   class BudFileReader < BudReadOnly # :nodoc: all
-    def initialize(name, filename, delimiter, bud_instance) # :nodoc: all
+    def initialize(name, filename, bud_instance) # :nodoc: all
       super(name, bud_instance, {[:lineno] => [:text]})
       @filename = filename
       @storage = {}

data/lib/bud/executor/elements.rb

@@ -199,7 +199,7 @@ module Bud
     public
     def pro(the_name=elem_name, the_schema=schema, &blk)
       toplevel = @bud_instance.toplevel
-      elem = Bud::PushElement.new("project#{object_id}",
+      elem = Bud::PushElement.new("project#{object_id}".to_sym,
                                   toplevel.this_rule_context,
                                   @collection_name, the_schema)
       self.wire_to(elem)
@@ -213,7 +213,7 @@ module Bud
     public
     def each_with_index(&blk)
       toplevel = @bud_instance.toplevel
-      elem = Bud::PushEachWithIndex.new("each_with_index#{object_id}",
+      elem = Bud::PushEachWithIndex.new("each_with_index#{object_id}".to_sym,
                                         toplevel.this_rule_context,
                                         @collection_name)
       elem.set_block(&blk)
@@ -284,7 +284,7 @@ module Bud
 
       aggpairs = prep_aggpairs(aggpairs)
       toplevel = @bud_instance.toplevel
-      g = Bud::PushGroup.new('grp'+Time.new.tv_usec.to_s, toplevel.this_rule_context,
+      g = Bud::PushGroup.new("grp#{Time.new.tv_usec}".to_sym, toplevel.this_rule_context,
                              @collection_name, keycols, aggpairs, the_schema, &blk)
       self.wire_to(g)
       toplevel.push_elems[[self.object_id, :group, keycols, aggpairs, blk]] = g
@@ -302,7 +302,7 @@ module Bud
       end
 
       aggpairs = [[agg, collection]]
-      aa = Bud::PushArgAgg.new('argagg'+Time.new.tv_usec.to_s, toplevel.this_rule_context,
+      aa = Bud::PushArgAgg.new("argagg#{Time.new.tv_usec}".to_sym, toplevel.this_rule_context,
                                @collection_name, gbkey_cols, aggpairs, schema, &blk)
       self.wire_to(aa)
       toplevel.push_elems[[self.object_id, :argagg, gbkey_cols, aggpairs, blk]] = aa
@@ -346,7 +346,7 @@ module Bud
     end
 
     def reduce(initial, &blk)
-      retval = Bud::PushReduce.new("reduce#{Time.new.tv_usec}",
+      retval = Bud::PushReduce.new("reduce#{Time.new.tv_usec}".to_sym,
                                    @bud_instance, @collection_name,
                                    schema, initial, &blk)
       self.wire_to(retval)
@@ -498,7 +498,7 @@ module Bud
       end
 
       # send deltas out in all cases
-      @collection.delta.each_value {|item| push_out(item)}
+      @collection.each_delta {|item| push_out(item)}
     end
   end
 

data/lib/bud/executor/join.rb

@@ -1,6 +1,5 @@
 require 'bud/executor/elements'
 
-$EMPTY = []
 module Bud
   class PushSHJoin < PushStatefulElement
     attr_reader :all_rels_below, :origpreds, :relnames, :keys, :localpreds
@@ -41,18 +40,6 @@ module Bud
         @selfjoins << name if cnt == 2
       end
 
-      # derive schema: one column for each table.
-      # duplicated inputs get distinguishing numeral
-      @cols = []
-      retval = @all_rels_below.reduce({}) do |memo, r|
-        r_name = r.qualified_tabname.to_s
-        memo[r_name] ||= 0
-        newstr = r_name + (memo[r_name] > 0 ? "_#{memo[r_name]}" : "")
-        @cols << newstr.to_sym
-        memo[r_name] += 1
-        memo
-      end
-
       setup_preds(preds) unless preds.empty?
       setup_state
 
@@ -143,14 +130,6 @@ module Bud
         if source_elem.rescan
           puts "#{qualified_tabname} rel:#{i}(#{source_elem.qualified_tabname}) invalidated" if $BUD_DEBUG
           @hash_tables[i] = {}
-          if i == 0
-            # Only if i == 0 because outer joins in Bloom are left outer joins.
-            # If i == 1, missing_keys will be corrected when items are populated
-            # in the rhs fork.
-            # XXX This is not modular. We are doing invalidation work for outer
-            # joins, which is part of a separate module PushSHOuterJoin.
-            @missing_keys.clear
-          end
         end
       end
     end
@@ -165,7 +144,7 @@ module Bud
       # referenced in entry.
       subtuple = 0
       all_rels_below[0..all_rels_below.length-1].each_with_index do |t,i|
-        if t.qualified_tabname == entry[0]
+        if t.qualified_tabname == name
           subtuple = i
           break
         end
@@ -183,7 +162,7 @@ module Bud
           elsif k.class <= Array
             [k,v]
           elsif k.class <= Symbol
-            if @all_rels_below and @all_rels_below.length == 2
+            if @all_rels_below.length == 2
               [find_attr_match(k, @all_rels_below[0]), find_attr_match(v, @all_rels_below[1])]
             else
               [find_attr_match(k), find_attr_match(v)]
@@ -235,9 +214,10 @@ module Bud
 
     protected
     def canonicalize_localpreds(rel_list, preds) # :nodoc:all
-      retval = preds.map do |p|
-        # reverse if lhs is rel_list[1], *unless* it's a self-join!
-        (p[0][0] == rel_list[1].qualified_tabname and p[0][0] != p[1][0]) ? p.reverse : p
+      second_rel = rel_list[1].qualified_tabname
+      preds.map do |p|
+        # reverse if lhs is second_rel *unless* it's a self-join!
+        (p[0][0] == second_rel and p[0][0] != p[1][0]) ? p.reverse : p
       end
     end
 
@@ -251,16 +231,15 @@ module Bud
       # again if we didn't rescan now.
       replay_join if @rescan
 
-      if @selfjoins.include? source.qualified_tabname
+      source_tbl = source.qualified_tabname
+      if @selfjoins.include? source_tbl
         offsets = []
-        @relnames.each_with_index{|r,i| offsets << i if r == source.qualified_tabname}
+        @relnames.each_with_index{|r,i| offsets << i if r == source_tbl}
       else
-        offsets = [@relnames.index(source.qualified_tabname)]
-      end
-      raise Bud::Error, "item #{item.inspect} inserted into join from unknown source #{source.elem_name}" if offsets == $EMPTY
-      offsets.each do |offset|
-        insert_item(item, offset)
+        offsets = [@relnames.index(source_tbl)]
       end
+
+      offsets.each {|offset| insert_item(item, offset)}
     end
 
     protected
@@ -332,14 +311,32 @@ module Bud
     ####
     # and now, the Bloom-facing methods
     # given a * expression over n collections, form all combinations of items
-    # subject to an array of predicates, pred
-    # currently supports two options for equijoin predicates:
+    # subject to an array of predicates, +preds+.
+    # currently supports two syntax options for equijoin predicates:
     #    general form: an array of arrays capturing a conjunction of equiv. classes
     #          [[table1.col1, table2.col2, table3.col3], [table1.col2, table2.col3]]
     #    common form: a hash capturing equality of a column on left with one on right.
     #          :col1 => :col2  (same as  lefttable.col1 => righttable.col2)
     public
     def pairs(*preds, &blk)
+      if @cols.nil?
+        # derive schema if needed: one column for each table.  duplicated inputs
+        # get distinguishing numeral.
+        #
+        # XXX: actually, this seems completely bogus. The schema for the output
+        # of the join should depend on the join's *targetlist*.
+        @cols = []
+        retval = @all_rels_below.reduce({}) do |memo, r|
+          r_name = r.qualified_tabname.to_s
+          memo[r_name] ||= 0
+          newstr = r_name + (memo[r_name] > 0 ? "_#{memo[r_name]}" : "")
+          @cols << newstr.to_sym
+          memo[r_name] += 1
+          memo
+        end
+        setup_accessors
+      end
+
       @origpreds = preds
       setup_preds(preds) unless preds.empty?
       # given new preds, the state for the join will be different.  set it up again.
@@ -361,20 +358,24 @@ module Bud
     end
 
     public
-    def rights(*preds, &blk)
-      @cols = blk.nil? ? @bud_instance.toplevel.tables[@rels[1].qualified_tabname].cols : nil
-      setup_accessors if blk.nil?
+    def lefts(*preds, &blk)
+      if blk.nil?
+        @cols = @bud_instance.toplevel.tables[@rels[0].qualified_tabname].cols
+        setup_accessors
+      end
       pairs(*preds) do |x,y|
-        blk.nil? ? y : blk.call(y)
+        blk.nil? ? x : blk.call(x)
       end
     end
 
     public
-    def lefts(*preds, &blk)
-      @cols = blk.nil? ? @bud_instance.toplevel.tables[@rels[0].qualified_tabname].cols : nil
-      setup_accessors if blk.nil?
+    def rights(*preds, &blk)
+      if blk.nil?
+        @cols = @bud_instance.toplevel.tables[@rels[1].qualified_tabname].cols
+        setup_accessors
+      end
       pairs(*preds) do |x,y|
-        blk.nil? ? x : blk.call(x)
+        blk.nil? ? y : blk.call(y)
       end
     end
 
@@ -405,13 +406,13 @@ module Bud
     public
     def flatten(*preds, &blk)
       if blk.nil?
-        @cols = dupfree_schema(@bud_instance.tables[@cols[0]].cols + @bud_instance.tables[@cols[1]].cols)
+        @cols = dupfree_schema(@rels[0].cols + @rels[1].cols)
       else
         @cols = []
       end
       setup_accessors
       pairs(*preds) do |x,y|
-        blk.nil? ? x.to_a + y.to_a : blk.call(x.to_a + y.to_a)
+        blk.nil? ? x + y : blk.call(x + y)
       end
     end
 
@@ -484,6 +485,14 @@ module Bud
         end
       end
     end
+
+    public
+    def invalidate_cache
+      super
+      # Only if need to check left join rel because outer joins in Bloom are
+      # left outer joins.
+      @missing_keys.clear if @rels.first.rescan
+    end
   end
 
 
@@ -497,11 +506,11 @@ module Bud
   # first flush, at which point we are sure to have seen all the t-side tuples
   # in this tick.
   class PushNotIn < PushStatefulElement
-    def initialize(rellist, bud_instance, preds=nil, &blk) # :nodoc: all
+    def initialize(rellist, bud_instance, preds, &blk) # :nodoc: all
       @lhs, @rhs = rellist
       @lhs_keycols = nil
       @rhs_keycols = nil
-      name_in = "#{@lhs.qualified_tabname}_notin_#{@rhs.qualified_tabname}"
+      name_in = "#{@lhs.qualified_tabname}_notin_#{@rhs.qualified_tabname}".to_sym
       super(name_in, bud_instance, nil, @lhs.schema)
       setup_preds(preds) unless preds.empty?
       @rhs_rcvd = false
@@ -532,12 +541,13 @@ module Bud
     end
 
     def find_col(colspec, rel)
-      if colspec.is_a? Symbol
+      case colspec
+      when Symbol
         unless rel.respond_to? colspec
           raise Bud::Error, "attribute :#{colspec} not found in #{rel.qualified_tabname}"
         end
         col_desc = rel.send(colspec)
-      elsif colspec.is_a? Array
+      when Array
         col_desc = colspec
       else
         raise Bud::Error, "symbol or column spec expected. Got #{colspec}"
@@ -546,8 +556,8 @@ module Bud
     end
 
     def get_key(item, offset)
-      keycols = offset == 0 ? @lhs_keycols : @rhs_keycols
-      keycols.nil? ? $EMPTY : item.values_at(*keycols)
+      keycols = (offset == 0 ? @lhs_keycols : @rhs_keycols)
+      keycols.nil? ? [] : item.values_at(*keycols)
     end
 
     public
@@ -580,8 +590,9 @@ module Bud
       # growing any more, until the next tick.
       unless @rhs_rcvd
         @rhs_rcvd = true
+        rhs_hash = @hash_tables[1]
         @hash_tables[0].each do |key,values|
-          rhs_values = @hash_tables[1][key]
+          rhs_values = rhs_hash[key]
           values.each {|item| process_match(item, rhs_values)}
         end
       end

data/lib/bud/lattice-core.rb

@@ -500,6 +500,11 @@ class Bud::LatticeWrapper
     end
   end
 
+  superator "<~" do |o|
+    # Overridden when <~ is defined (i.e., channels and terminals)
+    raise Bud::CompileError, "#{tabname} cannot appear on the lhs of a <~ operator"
+  end
+
   # XXX: refactor with BudCollection to avoid duplication of code
   def add_merge_target
     toplevel = @bud_instance.toplevel

data/lib/bud/monkeypatch.rb

@@ -10,17 +10,27 @@ class Class
   end
 end
 
+$struct_classes = {}
+$struct_lock = Mutex.new
+
 # FIXME: Should likely override #hash and #eql? as well.
 class Bud::TupleStruct < Struct
   include Comparable
 
+  def self.new_struct(cols)
+    $struct_lock.synchronize {
+      ($struct_classes[cols] ||= Bud::TupleStruct.new(*cols))
+    }
+  end
+
+  # XXX: This only considers two TupleStruct instances to be equal if they have
+  # the same schema (column names) AND the same contents; unclear if structural
+  # equality (consider only values, not column names) would be better.
   def <=>(o)
     if o.class == self.class
       self.each_with_index do |e, i|
         other = o[i]
         next if e == other
-        return nil if e.nil?
-        return nil if other.nil?
         return e <=> other
       end
       return 0
@@ -35,18 +45,27 @@ class Bud::TupleStruct < Struct
     if o.class == self.class
       return super
     elsif o.class == Array
-      begin
-        self.each_with_index do |el, i|
-          return false if el != o[i]
-        end
-        return true
-      rescue StandardError
-        return false
+      return false if self.length != o.length
+      self.each_with_index do |el, i|
+        return false if el != o[i]
       end
+      return true
     end
     false
   end
 
+  def hash
+    self.values.hash
+  end
+
+  def eql?(o)
+    self == o
+  end
+
+  def +(o)
+    self.to_ary + o.to_ary
+  end
+
   def to_msgpack(out=nil)
     self.to_a.to_msgpack(out)
   end
@@ -56,15 +75,23 @@ class Bud::TupleStruct < Struct
   end
 
   alias :to_s :inspect
+  alias :to_ary :to_a
 end
 
 # XXX: TEMPORARY/UGLY hack to ensure that arrays and structs compare. This can be
 # removed once tests are rewritten.
 class Array
-  alias :oldeq :==
+  alias :old_eq :==
+  alias :old_eql? :eql?
+
   def ==(o)
     o = o.to_a if o.kind_of? Bud::TupleStruct
-    self.oldeq(o)
+    self.old_eq(o)
+  end
+
+  def eql?(o)
+    o = o.to_a if o.kind_of? Bud::TupleStruct
+    self.old_eql?(o)
   end
 end