Opal does not support some language/runtime features of ruby. These are documented here when possible, as well as the reasons why they are not supported.
For performance and ease of runtime features, all strings in Opal are immutable, i.e. #<<, #gsub!, etc. do not exist. Also, symbols are just strings. There is no class, runtime or feature difference between Symbols and Strings. Their syntaxes can be used interchangeably.
We are using JavaScript regular expressions. While we do translate a few of Ruby specific instructions like \A or \z, there are a lot of incompatibilities that you should be aware of - for example $ and ^ don't match newlines like they do in Ruby. Support for features like named matches or lookahead/lookbehind is dependent on the JavaScript environment support for those. To support everything, we would need to compile in the entire Ruby's regular expression engine which is unfeasible at the current time.
In Opal, both integers and floats belong to same class Number (using JavaScript native numbers). So 1 / 4 is 0.25 (not 0) and 4.0 / 2 is 2 (not 2.0).
Number is a subclass of Numeric, like Complex and Rational.
Encodings only have a very small implementation inside Opal.
JavaScript does not have a native Thread implementation, so they are not present inside Opal. There is a placeholder Thread class just to provide some small level of compatibility with libraries that expect it. It does not have any function.
Opal does not currently support frozen objects, but has placeholder methods to prevent other libraries breaking when expecting these methods. Opal could support frozen objects in the future once a similar implementation becomes available across JavaScript runtimes.
All methods in Opal are defined as public to avoid additional runtime overhead. Module#private and Module#protected exist as just placeholder methods and are no-op methods.