stevegeek / encoded_id Goto Github PK

Add a new option to ReversibleId#initialize which specifies the maximum length an encoded ID can be (max_length)

The option should have a default value (say 128).

The option should be used in #encode, #decode, #encode_hex and #decode_hex.

On encoding, this limit value is used to check the generated encoded_id size before returning it.
If the generated encoded_id length is greater than max_length, raise a EncodedId::EncodedIdLengthError error

On decoding this limit should be used to check the input string before performing any operations.
If the input length is greater than max_length, raise a EncodedId::InvalidInputError error

The errors subclasses are all defined in lib/encoded_id.rb.

Note that a max_length of nil should result in no checks being made against the encoded_id and input lengths.

Please also add to the README.md to describe the option.

Add a new option to ReversibleId#initialize which specifies the maximum number of IDs that be encoded into any one encoded_id (max_inputs_per_id)

The option should have a default value (say 32).

The option should be used in #encode, & #encode_hex

If the input size is greater than max_inputs_per_id then raise InvalidInputError with an appropriate message

Sqids replace hashids , see https://sqids.org/ruby but are not backwards compatible

Upgrading to a v2 will require using a prefix on any new IDs to allow you to differentiate and if you have hash ID type ids in the wild, you will need to run both side by side and pick the right decoder... maybe the library can do this for you, ie be configured to support decoding older hashids (eg by identifying them someway, eg by format)

For perf reasons, should hashids ruby implementation be replaced with a rust based one? https://github.com/archer884/harsh

Or with YJIT no need and better to just use Ruby implementation?

Can the Ruby implementation be optimised with YJIT to reach Rust levels of perf?

The code behind this experimental feature needs to be refactored