* Add performance benchmark similar to ActiveRecord
* Lazily find_or_create_resource_for_collection to not incur the overhead for empty arrays and arrays of primatives
* #duplicable? is faster than inline rescues when the object is not duplicable
* Don't constantly raise and handle NameError, raising is expensive
* Even when a resource is nested inside a module, always look inside the class first for the resource definition so we don't overwrite classes all the time
Before: user system total real
Model.new (instantiation) 0.120000 0.000000 0.120000 ( 0.119961)
Nested::Model.new (instantiation) 0.150000 0.010000 0.160000 ( 0.151183)
Model.new (setting attributes) 28.540000 0.680000 29.220000 ( 29.271775)
Nested::Model.new (setting attributes) 29.740000 0.580000 30.320000 ( 30.486210)
After: user system total real
Model.new (instantiation) 0.120000 0.000000 0.120000 ( 0.121249)
Nested::Model.new (instantiation) 0.150000 0.010000 0.160000 ( 0.152429)
Model.new (setting attributes) 11.480000 0.170000 11.650000 ( 11.656163)
Nested::Model.new (setting attributes) 11.510000 0.210000 11.720000 ( 11.724249)
After a long list of discussion about the performance problem from using varargs and the reason that we can't find a great pair for it, it would be best to remove support for it for now.
It will come back if we can find a good pair for it. For now, Bon Voyage, `#among?`.
Following namespace use case was broken with Ruby 1.9:
class Author < ActiveRecord::Base
...
end
module Api
class Book < ActiveResouce::Base
end
end
Let's say XML contains <book><author><name>John</name></author>....
Api::Book.first.author.class.to_s #=>
Ruby 1.8.7: "Api::Book::Author" (namespaced, correct),
Ruby 1.9: "Author" (toplevel, broken)
Signed-off-by: José Valim <jose.valim@gmail.com>
Previously, ActiveResource was using the connection level formatter for get requests. This made it impossible to use custom formatters per resource.
Additionally this commit makes the Connection request methods more consistent. It always returns a Response. The base will then decode it each the response using its format setting.
Merging this commit will allow users to add custom formatters on a per Resource basis. This enables handling pagination responses from the server side, a very common use case that was previously impossible without monkeypatching XmlFormat.
Signed-off-by: José Valim <jose.valim@gmail.com>
This commit updates new? so that it knows whether or not the record was actually new or not, and doesn't rely solely on the presence of id. This enables the ability to set a custom primary_key that is not autogenerated by the server.
Signed-off-by: José Valim <jose.valim@gmail.com>
