rewriting most of the asn1 init code in ruby

[HoneyryderChuck]

This is an initial proposal towards the goal of having as much of the lib in ruby as possible. While most of the code must be in C, some of the code in C is essentially calling back to ruby, and writing "ruby in C", if that makes any sense.

Lemme know what you think. I understand there may also be some cognitive load from jumping from C to ruby and back, so there's certainly a trade-off.

[rhenium]

I'm all for it!

Apart from methods for accessing OpenSSL's OID registry (OpenSSL::ASN1::ObjectID.register, #sn, #ln, and #oid), OpenSSL::ASN1 could have been entirely written in Ruby, and that would definitely make maintenance easier.

It would be very nice if we can port the BER decoding code to Ruby.

[HoneyryderChuck]

@rhenium thx for the feedback!

Some follow-up where I need your input as well:

If you see the ruby code, you'll find the initialize method copy-pasted both for Primitive and Constructive, which is a direct port from the deleted C code, where both classes were sharing the same underlying impl in C. I'd like to address, and I can see the following options:

• Create base class for both, inheriting from ASN1Data, which implements common initialize;
• Create module with initialize, inject on both classes;
• Create __initialize_with.. method in ASN1Data class, call for both on initialize;

Essentially, the first 2 options will require a potential breaking(?) change in the ancestor chain. Not sure how you feel about that. The third option doesn't do that, at the cost of adding an additional "hidden" method. All of the above will add smth new to document (or maybe we can :nodoc: all of them?). Which one do you think suits this the best?

It would be very nice if we can port the BER decoding code to Ruby.

You mean OpenSSL::ASN1.decode? I can try, but AFAIR there was some dependency on some openssl function calls, so don't know how feasible will that be.