You can now graph a dataset and have the result split into component tables:
DB[:artists].graph(:albums, :artist_id=>:id).first # => {:artists=>{:id=>artists.id, :name=>artists.name}, \ # :albums=>{:id=>albums.id, :name=>albums.name,
:artist_id=>albums.artist_id}}
This aliases columns if necessary so they don’t stomp on each other, which is what usually happens if you just join the tables:
DB[:artists].left_outer_join(:albums, :artist_id=>:id).first # => {:id=>(albums.id||artists.id),
:name=>(albums.name||artist.names), \
:artist_id=>albums.artist_id}
Models can use graph as well, in which case the values will be model objects:
Artist.graph(Album, :artist_id=>:id) # => {:artists=>#<Artist...>, :albums=>#<Album...>}
Models can now eager load via .eager_graph, which will load all the results and all associations in a single query. This is necessary if you want to filter on columns in associated tables. It works exactly the same way as .eager, and supports cascading of associations as well:
# Artist.one_to_many :albums # Album.one_to_many :tracks # Track.many_to_one :genre Artist.eager_graph(:albums=>{:tracks=>:genre}).filter( \ :tracks_name=>"Firewire").all
This will give you all artists have have an album with a track named “Firewire”, and calling .albums on one of those artists will only return albums that have a track named “Firewire”, and calling .tracks on one of those albums will return only the track(s) named “Firewire”.
You can use set_graph_aliases to select specific columns:
DB[:artists].graph(:albums, :artist_id=>:id).set_graph_aliases( \ :artist_name=>[:artists, :name], :album_name=>[:albums,
:name]).first
# => {:artists=>{:name=>artists.name}, :albums=>{:name=>albums.name}}
You can use eager_graph with set_graph_aliases to have eager loading with control over the SELECT clause.
All associations now update their reciprocal associations whenever the association methods are used, so you don’t need to refresh the association or model to have the reciprocal association updated:
Album.many_to_one :band Band.one_to_many :albums # Note that all of these associations are cached, # so after the first access there are no additional # database queries to fetch associated records. # many_to_one setter adds to reciprocal association band1.albums # => [] album1.band = band1 band1.albums # => [album1] band2.albums # => [] album1.band = band2 band1.albums # => [] band2.albums # => [album1] album1.band = band2 band2.albums # => [album1] album1.band = nil band2.albums # => [] # one_to_many add_* method sets reciprocal association # one_to_many remove_* method removes reciprocal association album1.band # => nil band1.add_album(album1) album1.band # => band1 band2.add_album(album1) album1.band # => band2 band2.remove_album(album1) album1.band # => nil Post.many_to_many :tags Tag.many_to_many :posts # many_to_many add_* method adds to reciprocal association # many_to_many remove_* method removes from reciprocal association post1.tags # => [] tag1.posts # => [] tag1.add_post(post1) post1.tags # => [tag1] tag1.posts # => [post1] tag1.remove_post(post1) post1.tags # => [] tag1.posts # => [] post1.add_tag(tag1) post1.tags # => [tag1] tag1.posts # => [post1] post1.remove_tag(tag1) post1.tags # => [] tag1.posts # => []
The MySQL and PostgreSQL adapters now support index types:
index :some_column, :type => :hash # or :spatial, :full_text, :rtree,
etc.
Starting in Sequel 1.5.0, some methods are deprecated. These methods will be removed in Sequel 2.0.0. The deprecation framework is fairly flexible. You can choose where the messages get sent:
Sequel::Deprecation.deprecation_message_stream = STDERR # the default Sequel::Deprecation.deprecation_message_stream = \ File.new('deprecation.txt', 'wb') # A file Sequel::Deprecation.deprecation_message_stream = nil # ignore the
messages
You can even have all deprecation messages accompanied by a traceback, so you can see exactly where in your code you are using a deprecated method:
Sequel::Deprecation.print_tracebacks = true
All deprecation methods come with an message telling you what alternative code will work.
In addition to deprecating some methods, we removed the ability to have arrays returned instead of hashes. The array code still had debugging messages left it in, and we are not aware of anyone using it. Hashes have been returned by default since Sequel 0.3.
We have also removed the Numeric date/time extensions (e.g. 3.days.ago). The existing extensions were incomplete, better ones are provided elsewhere, and the extensions were not really related to Sequel’s purpose.
Sequel no longer depends on ParseTree, RubyInline, or ruby2ruby. They are still required to use the block filters. Sequel’s only gem dependency is on the tiny metaid.
Sequel 1.5.0 has fixes for 12 tracker issues, including fixes to the Informix, MySQL, ODBC, ADO, JDBC, Postgres, and SQLite adapters.