So I just started an empty rails project this morning and tested it all out. Here's the steps to reproduce all errors.

1. rails new project with ruby 3.2.0 and rails 7.0.8.
2. Add to Gemfile: (also tried w/out 'ksuid' same error)

gem 'ksuid'
gem 'activerecord-ksuid', require: 'active_record/ksuid/railtie'

3. Run rails generate model Event my_field_name:ksuid => creates error.
4. Create a migration file and use add_column => creates error.
5. Create a migration file and use create_table => does not create error and works as expected.

Originally posted by @DataDaoDe in #21 (comment)

From the README:

Then, to add proper handling to the field, you will want to mix a module into the model:

class Event < ApplicationRecord
  include KSUID::ActiveRecord[:unique_id]
end

Would it be possible to reflect the fields on a table and automate this step? That way we would not have to explicitly write include KSUID::ActiveRecord[:unique_id].

# Could not dump table "xxx" because of following StandardError
#   Unknown type 'ksuid' for column 'id'

Rails 6.0.2.1

Hey!

Not sure what information you need, but when trying to add the KSUID type in an existing modal, I get the following error:

rake db:migrate
/Users/XXXXXX/.rbenv/versions/2.7.1/lib/ruby/gems/2.7.0/gems/activerecord-6.0.3.2/lib/active_record/associations.rb:1370: warning: The called method `has_many' is defined here
== 20201215203500 AddKsuidToUserProjectDownload: migrating ====================
-- add_column(:user_project_downloads, :unique_id, :ksuid)
rake aborted!
StandardError: An error has occurred, this and all later migrations canceled:

PG::UndefinedObject: ERROR:  type "ksuid" does not exist
LINE 1: ALTER TABLE "user_project_downloads" ADD "unique_id" ksuid
                                                             ^
/Users/XXXXXX/db/migrate/20201215203500_add_ksuid_to_user_project_download.rb:3:in `change'

Caused by:
ActiveRecord::StatementInvalid: PG::UndefinedObject: ERROR:  type "ksuid" does not exist
LINE 1: ALTER TABLE "user_project_downloads" ADD "unique_id" ksuid
                                                             ^
/Users/XXXXXX/db/migrate/20201215203500_add_ksuid_to_user_project_download.rb:3:in `change'

Caused by:
PG::UndefinedObject: ERROR:  type "ksuid" does not exist
LINE 1: ALTER TABLE "user_project_downloads" ADD "unique_id" ksuid
                                                             ^
/Users/XXXXXX//db/migrate/20201215203500_add_ksuid_to_user_project_download.rb:3:in `change'
Tasks: TOP => db:migrate
(See full trace by running task with --trace)

With a migration that looks like this:

class AddKsuidToUserProjectDownload < ActiveRecord::Migration[6.0]
  def change
    add_column :user_project_downloads, :unique_id, :ksuid
  end
end

We're currently using SecureRandom by default, but some people might have different constraints for their pseudorandomness. As such, we should make the psuedorandom number generator pluggable.

We have been using include KSUID::ActiveRecord[:project_id] to declare KSUID attributes in models.

That's no problem for primary keys ("id"). For foreign keys this leads to the attribute initialized with a default KSUID, because of the default. OK and necessary for primary keys; not necessary and misleading for foreign keys.

I will no suggest to declare ksuid foreign keys with

attribute :project_id, :ksuid

instead.

Theoretically it would be nice if this could be derived from introspecting the relationship.

Or am I missing something?

One solution might be to have a separate concern for primary and foreign keys or to make setting the default optional somehow.

Per this guide, if one needs uuid support, both the active storage table and the record_id reference need to be of the uuid type.

How can I make this work with ksuid?

I'm guessing ksuid is not a type pg will recognize.

It also does not seem to have an impact when I append type: :ksuid to the references statement.

In the generated schema, the record_id is still a bigint.

Please advise

Hi,

the readme references adding the activerecord-ksuid gem to the Gemfile, but it can't be found on rubygems and looks like it's not released. Is this the case? Is it in a production ready state?

Thanks,
Daniel

From this request.

This should be straightforward:

1. During table definition for a string-serialized column, detect whether a prefix was passed and adjust the limit by the length of the prefix
2. During table definition for a binary-serialized column, throw an exception because it wouldn't make sense to have a prefix for a binary column
3. Extend the model mixin to handle prefixes. This will necessitate a new ActiveRecord::Type class that combines the prefix with the KSUID.
   a. The mixin should throw an error if you combine prefix: with binary: true because, again, that doesn't make sense.
   b. The new composite type probably will need some massaging to interface with created_at: true.
4. Write up documentation about it everything.

class House
  include KSUID::ActiveRecord[:id]
  has_many :people
end

class Person
  include KSUID::ActiveRecord[:id]
  belongs_to :house
end

Person.new(house: House.last)

This throws an error by default.

Libraries are a lot friendlier to use and work on when their public interface is documented. We should add documentation to the entire public interface for the gem. This should be done in the YARD format. We should then wire up the documentation helpers into the default rake task.

The highest impact guide would be using these in ActiveRecord, due to Rails' influence on the Ruby community. But it would also be interesting to see the following:

1. Sequel
2. ROM
3. Mongoid - This one is maybe not worth it since MongoDB already has their ObjectIDs.

On further reflection, I won't write a Mongoid integration without being asked (or preferably helped). I haven't used Mongoid in years and don't think I would do a good job of it. In addition, the aforementioned ability of their ObjectIDs semi obviates the need for KSUIDs in that context.

This happens when trying to eager load associations:

Organization.includes(:users).each do |o|
    ...
 end

Throws:

NoMethodError: undefined method `first' for nil:NilClass

The problem is in:
activerecord-6.1.4.1/lib/active_record/associations/preloader/association.rb:122:in block in records_for'`

Specifically, this line:

owner = owners_by_key[convert_key(record[association_key_name])].first

from

        def records_for(ids)
          scope.where(association_key_name => ids).load do |record|
            # Processing only the first owner
            # because the record is modified but not an owner

            owner = owners_by_key[convert_key(record[association_key_name])].first
            association = owner.association(reflection.name)
            association.set_inverse_instance(record)
          end
        end

In order to lookup associated records, Activerecord looks for the key stored as KSUID::Type. For instance, value of owners_by_key:

{<KSUID(1ym0xC8YYVoho8JmB19Y5Rsdi3O)>=>[#<Organization id: <KSUID(1ym0xC8YYVoho8JmB19Y5Rsdi3O)>...}

And the value for convert_key(record[association_key_name]):

<KSUID(1ym0xC8YYVoho8JmB19Y5Rsdi3O)>

It's the same id. But when trying to fetch that key the result is nil. I was assuming it had to do with how KSUID converts to object_id, as the two 'same' KSUIDs convert to different object_ids. Leading to not finding the key as the two object_ids are different.

I tried to define #object_id in KSUID::Type to return a static value:

    def object_id
      self.to_i
    end

However, that doesn't fix the issue and still, the key is not found in Activerecord Association's class' owners_by_key hash.

Any ideas how to fix this?

I started doing some compatibility tests and noticed that we're incompatible with the main library, even though the max is decoded properly. I have a feeling this is due to some endianness issues. My use of Array#pack and String#unpack is probably incorrect.

One thing I am finding that I do a lot of is build a new KSUID and immediately coerce it to a string.

Have you thought about a public interface for building a string directly? Is there any preferable nomenclature for this sort of thing?

Hello there!

My name is Ana. I noted that you use the mutation testing tool mutant in the project.
I am a postdoctoral researcher at the University of Seville (Spain), and my colleagues and I are studying how mutation testing tools are used in practice. With this aim in mind, we have analysed over 3,500 public GitHub repositories using mutation testing tools, including yours! This work has recently been published in a journal paper available at https://link.springer.com/content/pdf/10.1007/s10664-022-10177-8.pdf.

To complete this study, we are asking for your help to understand better how mutation testing is used in practice, please! We would be extremely grateful if you could contribute to this study by answering a brief survey of 21 simple questions (no more than 6 minutes). This is the link to the questionnaire https://forms.gle/FvXNrimWAsJYC1zB9.

Drop me an e-mail if you have any questions or comments ([email protected]). Thank you very much in advance!!

0.1.0 appears to be an old version that doesn't match the documentation for rails.