Participant required fields + CoC checkbox cleanup

[experimatt]

While testing for #337, I realized that neither email nor password were required fields on the create account page.

This cleans up a number of small things:

Participants

• Require email and password field to be present when creating an account
• Add a feature test for the above to ensure this continues to work as expected
• Add database NOT NULL constraints for email and crypted_password
  • I verified that there aren't currently any participants with null email or crypted_password values in the DB, but wrote the migration in a way that it shouldn't matter too much (e.g. null gets replaced with "INVALID" if any exist).
• Related ^: Remove Admin::Legacy::SessionsController#create and #new endpoints, along with problematic build_presenter method that would create a participant without an email or password, thus triggering these new constraints.
  • Given that we're moving away from the legacy admin interface, I thought ripping this code and tests out was fine. Plus it's not behavior we really want to support.

Code of Conduct Agreements

• Add has_many associations to Event and Participant for easy access
• Disable the CoC checkbox on the create/edit session form when it has already been accepted for the event.

Other small changes

• Fix email input field width so it matches the name and password fields on registration.
• Use Time.current instead of Time.now for email_confirmed_at, so it's not saved with a time zone.

Screen

Before

shots

After
This shows the wider email input + validation errors on email and password, which weren't present before.

[Copilot]

PR Overview

This PR reinforces validation on account creation by enforcing required email and password fields, introduces associations for Code of Conduct agreements, and cleans up related UI components and timestamp handling.

• Enforces email and password presence and uniqueness in the Participant model
• Adds feature tests to cover required field validations
• Updates associations and UI elements for Code of Conduct agreements and uses Time.current for timestamp consistency

Reviewed Changes

File	Description
spec/features/sign_in_and_out_spec.rb	Adds a feature test to verify the presence of required fields
app/models/participant.rb	Updates validations and association for Code of Conduct agreements
app/controllers/participants_controller.rb	Updates email confirmation timestamp handling to Time.current
app/models/event.rb	Adds association for Code of Conduct agreements
app/views/sessions/_form.html.erb	Modifies the CoC checkbox to be disabled if already accepted
app/views/participants/new.html.erb	Marks the CoC field as required (adds an asterisk to the label)

Copilot reviewed 7 out of 7 changed files in this pull request and generated 1 comment.

[pcantrell]

Good catch on the form allowing null emails! Ha. And yikes.

All the changes LGTM.

One suggestion: I’d add a migration that adds a not null constraint for email and crypted_password in the DB too, just to be sure. (No existing records appear to be null or blank.) In practice, nobody’s creating accounts except through the UI, but…a little DB quality diligence now may pay dividends in time.

[mkbaker]

this looks good to me too!

[experimatt]

One suggestion: I’d add a migration that adds a not null constraint for email and crypted_password in the DB too, just to be sure. (No existing records appear to be null or blank.) In practice, nobody’s creating accounts except through the UI, but…a little DB quality diligence now may pay dividends in time.

@pcantrell how's this look? I added the NOT NULL constraints you suggested, which caused some tests to fail and exposed some problematic code in the legacy admin sessions controller that I just ripped out entirely.

If this looks good to you, let's merge and deploy this bad boy.

[pcantrell]

Strictly speaking, migrations shouldn’t use models, because this migration has to work with every future version of the model code (but with the schema frozen at this point in the migration chain). In practice, however, I really doubt this will cause any actual problems!

[experimatt]

That's fair. I agree with you - I think the likelihood that the Participant model goes away are very low.

[pcantrell]

LGTM with the new changes!