Unexpected behaviour of contact code

[artur-intech]

I have discovered the following issue while testing EPP update: I manually created new contact with the code "test". Then I tried to update it via EPP by specifying "test" as a content of "id" tag, but it failed, since the code is saved to the database as is if created outside of EPP, but is converted to uppercase when making EPP query.

• It is unexpected
• Against least surprise principle
• Impossible to remember

Last thing leads to additional effort finding out why newly added contact is not found. It is also unnatural to always specify the code in uppercase. Consider this example https://github.com/internetee/registry/blob/master/spec/requests/epp/contact/update/phone_spec.rb: it opens up a question: why TEST is just upper-cased?!

Querying a contact via EPP fails if a contact is created using one of the following statements:

Contact.create!(name: 'test', code: 'test1', phone: '+123.4', email: 'test111@internet.ee', country_code: 'EE', ident: '37605030299', ident_type: 'priv', ident_country_code: 'EE', registrar: Registrar.last)

RSpec.describe 'EPP contact:update' do
  let!(:contact) { create(:contact, code: 'test') }
  let(:request_xml) { <<-XML
    <?xml version="1.0" encoding="UTF-8" standalone="no"?>
    <epp xmlns="https://epp.tld.ee/schema/epp-ee-1.0.xsd">
      <command>
        <update>
          <contact:update xmlns:contact="https://epp.tld.ee/schema/contact-ee-1.1.xsd">
            <contact:id>test</contact:id>
...

I consider this kind of behaviour harmful and suggest that we either remove auto uppercase in EPP at all or make it predictable by adding auto conversion to Contact model, so it is always saved upper-cased.

[vohmar]

conversion is a must.
You can move conversion to contact model if it helps the development but nothing must change for EPP users.

[artur-intech]

https://nandovieira.com/using-insensitive-case-columns-in-postgresql-with-citext