Elia Schito

> @elia You've made some changes to break behavior about a month ago. Can it be related? absolutely ☹️ especially if it used to work properly, otherwise could be one...

update: ``` opal:elia/bare-necessities ⤑ be ruby -e 'require "opal"; p size: Opal::Builder.new(optimize_calls: true).build("opal").to_s.size' ~/C/opal DCE.37.96.163.195.209.214.215.215 {:size=>419772} opal:elia/bare-necessities ⤑ be ruby -e 'require "opal"; p size: Opal::Builder.new(optimize_calls: false).build("opal").to_s.size' ~/C/opal {:size=>579802} ```...

It's intended as heavy optimization, so not supported, same goes for `method_missing`. An option to mitigate this limitation could be to restrict the DCE to the core/std lib, assuming you're...

It is, any possible improvement in performance is of secondary importance in this context (eg for not passing a TMP var for the method to retrieve the block)

Yes, from what I found it ignores the double parens on the lhs of the assignment: ``` opal:master ⤑ ruby -ve '(x, y) = [[1,2]]; p(["x", x, "y", y])' ~/C/opal...

Yeah, CWD has a shim in the browser just to make most gems happy: https://github.com/opal/opal/blob/master/opal/corelib/dir.rb and is linked to Node's implementation when you `require 'nodejs'` from the stdlib https://github.com/opal/opal/blob/master/stdlib/nodejs/dir.rb (disclaimer:...

The why is that any method from kernel can conflict with a native method, so we prefer to keep it as an opt-in option and avoid headaches down the road....

As a wrapper could live in stdlib or even in an external gem I think

@opal/owners seems reasonable to add bignum support to the compiler If that's not something we want to support it makes sense to raise an error when the parser encounters a...

I assume you mean for nodejs. If you can hint on which nodejs counterpart could be used, it would be great.