This reverts commit c24c8852.

Here's the explanation I just sent to @tenderlove:

Hey,

I've been thinking about about the transaction memory leak thing that we
were discussing.

Example code:

post = nil
Post.transaction do
  N.times { post = Post.create }
end

Post.transaction is going to create a real transaction and there will
also be a (savepoint) transaction inside each Post.create.

In an idea world, we'd like all but the last Post instance to be GC'd,
and for the last Post instance to receive its after_commit callback when
Post.transaction returns.

I can't see how this can work using your solution where the Post itself
holds a reference to the transaction it is in; when Post.transaction
returns, control does not switch to any of Post's instance methods, so
it can't trigger the callbacks itself.

What we really want is for the transaction itself to hold weak
references to the objects within the transaction. So those objects can
be GC'd, but if they are not GC'd then the transaction can iterate them
and execute their callbacks.

I've looked into WeakRef implementations that are available. On 1.9.3,
the stdlib weakref library is broken and we shouldn't use it.

There is a better implementation here:

https://github.com/bdurand/ref/blob/master/lib/ref/weak_reference/pure_ruby.rb

We could use that, either by pulling in the gem or just copying the code
in, but it still suffers from the limitation that it uses ObjectSpace
finalizers.

In my testing, this finalizers make GC quite expensive:
https://gist.github.com/3722432

Ruby 2.0 will have a native WeakRef implementation (via
ObjectSpace::WeakMap), hence won't be reliant on finalizers:
http://bugs.ruby-lang.org/issues/4168

So the ultimate solution will be for everyone to use Ruby 2.0, and for
us to just use ObjectSpace::WeakMap.

In the meantime, we have basically 3 options:

The first is to leave it as it is.

The second is to use a finalizer-based weakref implementation and take
the GC perf hit.

The final option is to store object ids rather than the actual objects.
Then use ObjectSpace._id2ref to deference the objects at the end of the
transaction, if they exist. This won't stop memory use growing within
the transaction, but it'll grow more slowly.

I benchmarked the performance of _id2ref this if the object does or does
not exist: https://gist.github.com/3722550

If it does exist it seems decent, but it's hugely more expensive if it
doesn't, probably because we have to do the rescue nil.

Probably most of the time the objects will exist. However the point of
doing this optimisation is to allow people to create a large number of
objects inside a transaction and have them be GC'd. So for that use
case, we'd be replacing one problem with another. I'm not sure which of
the two problems is worse.

My feeling is that we should just leave this for now and come back to it
when Ruby 2.0 is out.

I'm going to revert your commit because I can't see how it solves this.
Hope you don't mind... if I've misunderstood then let me know!

Jon

In some circumstances engine was Arel::Table.engine which for separate
reasons was an ActiveRecord::Model::DeprecationProxy, which caused a
deprecation warning.

In any case, we want the actual model class here, since we want to use
it to infer information about associations.

Previously the reflection would be looked up on the wrong class. However
the test passed because the examples referred back to themselves.

single-table inheritance by overriding it in your ActiveRecord
Model.

1. Unused variable
2. possibly useless use of a variable in 
void context

- added tests to confirm establish_connection uses DATABASE_URL and
  Rails.env correctly even when no arguments are passed in.
- updated rake db tasks to support DATABASE_URL, and added tests to
  confirm correct behavior for these rake tasks.  (Removed
  establish_connection call from some tasks since in those cases
  the :environment task already made sure the function would be called)
- updated Resolver so that when it resolves the database url, it
  removes hash values with empty strings from the config spec (e.g.
  to support connection to postgresql when no username is specified).

Allows you to specify the model association key in a belongs_to
relationship instead of the foreign key.

The following queries are now equivalent:

Post.where(:author_id => Author.first)
Post.where(:author => Author.first)

PriceEstimate.where(:estimate_of_type => 'Treasure', :estimate_of_id => treasure)
PriceEstimate.where(:estimate_of => treasure)

As a result of different commits, ConnectionPool had become
of two minds about exceptions, sometimes using PoolFullError
and sometimes using ConnectionTimeoutError. In fact, it was
using ConnectionTimeoutError internally, but then recueing
and re-raising as a PoolFullError.

There's no reason for this bifurcation, standardize on
ConnectionTimeoutError, which is the rails2 name and still
accurately describes semantics at this point.

History

In Rails2, ConnectionPool raises a ConnectionTimeoutError if
it can't get a connection within timeout.

Originally in master/rails3, @tenderlove had planned on removing
wait/blocking in connectionpool entirely, at that point he changed
exception to PoolFullError.

But then later wait/blocking came back, but exception remained
PoolFullError.

Then in 02b23355 pmahoney introduced fair waiting logic, and
brought back ConnectionTimeoutError, introducing the weird bifurcation.

ConnectionTimeoutError accurately describes semantics as of this
point, and is backwards compat with rails2, there's no reason
for PoolFullError to be introduced, and no reason for two
different exception types to be used internally, no reason
to rescue one and re-raise as another.  Unify!

Exception message was misleading, as it is possible to have a
polymorphic 'has_many :through' join model.

When calling a query method on an attribute that was not selected by
an ActiveRecord query, an ActiveModel::MissingAttributeError is not
raised. Instead, a nil value is returned, which will return false once
cast to boolean.

This is undesirable, as we should not give the impression that we know
the attribute's boolean value when we haven't loaded the attribute's
(possibly) non-boolean value from the database.

This issue is present on versions going back as far as 2.3, at least.

Merged in f41dba27
[ci skip]

* There is no need to delete the primary key from cloned attributes,
  since it sets the same pk to nil afterwards.
* Check for empty? instead of any? to run initialize callbacks.

transaction.

Check 0180e090 for more reasoning about
that.

Since 810a50da, the new policy is to
keep old changelogs in their own branches, to avoid manual syncing
across different branches.

Please check that commit for more reasoning about the new policy.

move validation to AR

This will solve the issue that abort the connection transaction when we
skip the tests.

This implements the support to encode/decode JSON
data to/from database and creating columns of type
JSON using a native type [1] supported by PostgreSQL
from version 9.2.

[1] http://www.postgresql.org/docs/9.2/static/datatype-json.html

The previous implementation had the strange requirement that db/structure.sql contain only CREATE TABLE sql statements, one per table, separated by double newlines. SQLite3 and PostgreSQL database tasks, on the other hand, simply spawn 'sqlite3' and 'psql' binaries to load the file directly. The new implementation follows this and attempts to respect all current MySQL configuration settings.