Prior to this patch every time the schema was regenerated at the end of
the file it prepended an additional newline character. This made the
schema slowly creep down the file over time. Finally annoyed me enough
to fork and fix ;)

Recognize column+type, and don't change a file unless the column+type combination of the new schema are different than that of the old (i.e., don't regenerate if columns happen to be in a different order. That's just how life is sometimes)
Grab old specification even if it has \r\n as line endings rather than pure \ns
Remove gemspec line causing a warning

model file, it will be replaced into the same location.  If it didn't
previously exist, it'll be placed according to the "position", as before.

added '-R' option to require additional files
ignore unknown macros ('acts_as_whatever') (with specs!)

use model_dir parameter in remove_annotation as well; remove obsolete (and confusing) annotate.rake file

fix the dependencies so it doesn't fail when multiple version of Rails is installed, reported by kuroda

A class named Approval in moderations/approval.rb raises no error
Signed-off-by: ctran <ctran@pragmaquest.com>

Before this change only default(TRUE) was written.

Signed-off-by: ctran <ctran@pragmaquest.com>

Fix a bug where options are not passed when annotating tests/specs.

tagging 2.2.2

Adds compat mode for annotate plugin version
Tim Charper binary write mode back
Updates history

* update gem with newgem
Add my plugin addons:
* annotate geometrical columns
* annotate object daddy
* annotate spec models
Annotates in the end by default.

Modified the AnnotateModels module so it only updates models that have changed since the last time 'annotate' was run. This is huge for those of us who like to run 'annotate' frequently but don't want to update our VCS for all the models in our application each time we do.

This required a few small modifications to the annotate_one_file, annotate, and do_annotations methods. Comments have been updated to reflect changes.

= In ruby, on windows =
* A text file with CRLF or LF line endings will be opened with UNIX (LF) line endings in ruby.
* When writing text to a file, ruby always uses CRLF line endings, regardless of the format of the line endings of the file to begin with.

This causes an automatic conversion of all LF files to CRLF.  If you check out a project that uses unix line endings, then run annotate_models (before this commit), you're SCM is going to want to commit every line of every model.

= In ruby, on *nix / osx =
- Strings are always read and written as text with LF line endings.

----

The end result is an endless tug-a-war over line-endings for collaborating windows and *nix developers.

If you specify to write the files as binary ("wb"), both windows and *nix will output the text with LF line endings.  Most editors on windows support this format, and if you're deploying to a linux server, you should be using this format anyways.  This has solved the tug-a-war for me.