opengraph_parser is basically unmainted, issues are ignored
or deliberately closed without fixing. It pollutes the global
namespace and has no verification of correctness.
The opengraph gem has basically the same issues, not really maintained,
unreleased patches on master since over a year, not really smart either.
So I created my own version and while at it, why not strive try to be
complete and robust, although it's still a work in progress.
This also improves general URL detection by parsing them
from the message after stripping markdown.
An additional dependency was added to support
fetching sites that require cookies to work at all.
For the same reason Faraday's default redirect limit was
bumped.
* Get rid of early pathname requirement
* Use require_relative where approciate
* Drop unused files
* Imported new application.rb, environment.rb and environments/* from fresh Rails app
* Cleaned up boot.rb
* Load config in boot.rb
* Deduplicate environments/integration*.rb
* Move username blacklist into defaults.yml
* Ruby 1.9 Hash syntax everywhere
* Reorganize lib/diaspora/markdownify to match conventions
* Get rid of full path requires where possible
* Add dummy content to production section diaspora.yml.example to prevent warning
* Drop sqlite? method
* Move postgres? method into ConfigurationMethods
* Drop token authentication from User
* Throw away old system
* Add new system
* Add new example files
* Replace all calls
* add the most important docs
* Add Specs
* rename disable_ssl_requirement to require_ssl
* cloudfiles isn't used/called in our code
* since community_spotlight.list is only used as enable flag replace it with such one and remove all legacy and irelevant codepaths around it
* die if session secret is unset and on heroku
* First basic infrastructure for version information
