norman / ambry Goto Github PK

I'm working on a gem that would baiscally wrap lot of yaml data files using ambry, but these shouldn't be modifiable

Should an ambry model be subclassed with following syntax or similar?:

require 'ambry'
require 'ambry/active_model'

class MyAppBadge < Badge
  extend Ambry::Model
  extend Ambry::ActiveModel

  field :type
end

Right now it doesn't work like that:

Badge.new.attributes # => {"id"=>nil, "name"=>nil, "level"=>nil, "image"=>nil, "description"=>nil} 
MyAppBadge.new.attributes # => {"type"=>nil} 

But following code seems to play nicer:

require 'ambry'
require 'ambry/active_model'

class MyAppBadge < Badge
  extend Ambry::Model
  extend Ambry::ActiveModel

  field *Badge.new.attributes.keys, :type
end
Badge.new.attributes.keys # => {"id"=>nil, "name"=>nil, "level"=>nil, "image"=>nil, "description"=>nil} 
MyAppBadge.new.attributes # => {"id"=>nil, "name"=>nil, "level"=>nil, "image"=>nil, "description"=>nil, "type"=>nil} 

See also merit-gem/merit#28.

Thanks!

Tute.

I think a default in-memory adapter should be implied, without the need to explicitly create one

Ambry uses blank? in few places, but has no activesupport dependency (only in development), which results in method missing exceptions. Should we add activesupport dependency or simply change few instances of blank? to nil? || empty??

Currently, if we have the following model

class Person
  extend Ambry::Model
  field :email, :name
end

when initialized with following yaml data

[email protected]:
  name: Foo

the "email" field won't be set, but is needed to be explicitly present:

[email protected]:
  name: Foo
  email: [email protected]