Skip to content
This repository has been archived by the owner on Oct 26, 2023. It is now read-only.

raise PersistenceError if invalid values #1

Closed
wants to merge 73 commits into from

Conversation

lanej
Copy link

@lanej lanej commented Aug 17, 2015

  • currently fails silently and without anyway of knowing why

Markus Schirp and others added 30 commits May 29, 2011 22:49
The intention is to use the shared adapter specs from
dm-mongo-adapter without the need to duplicate it.

Tested against sqlite-adapter mongo-adapter and in-memory adapter, on
ruby-1.9.2-p180.
Conflicts:
	lib/dm-core/property/binary.rb
Remove jeweler dev dep and clean up gemspec/gemfile
* In general command methods like finalize will return self, as the DM API
  moves closer towards using CQS like Veritas, Virtus and other libraries
  that will be part of DM2.
dkubb and others added 29 commits February 14, 2012 22:11
* In Virtus the string "yes" is now coerced to true, causing the spec to fail
* These assertions were originally meant to prevent common errors, but I now
  believe they do more harm than good because they prevent other kinds of
  objects from being injected into the system and then used in the adapters.
Support both the ADAPTER and ADAPTERS environment variables.
* Usually my convention when subclassing is to use the same visibility on any
  methods I'm overriding. Since all objects in Ruby inherit from Object, and
  Object#const_missing is public, then my own #const_missing methods should also
  be public.
DataMapper::Hook .method_added and .singleton_method_added should call super
* All callback methods (extended, included, inherited, etc) should call super
  when they execute, even if it's likely to be a noop, because you can't assume
  that an ancestor does not have a same-named method defined.

  There are also cases (which are a bit dubious) where core classes have been
  monkey patched and had these methods added, and if you don't call super then
  they will never execute. While I don't necessarily agree with this practice, I
  feel like *not* calling super is more wrong. If developers want to do this I
  need to assume they are willign to deal with the consequences.
* Models can be bound to specific repositories, but we want for the tests to
  always auto-migrate the models within the specified repository.
* If a model was created for some other purpose, besides testing its
  persistence logic, it's likely it was an anonymous class and may be considered
  incomplete and impossible to auto-migrate.
This makes it possible to have dm-core coexist
with veritas and friends.

virtus 0.5 hasn't yet been released so virtus is
pinned to github in the Gemfile for now.
correct singularization of status (and alias)
Change to support multiple threads in jruby--fixes github issue datamapper#188
@lanej lanej closed this Aug 17, 2015
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
9 participants