Decisions concerning the Grader

[tgarg10]

👋 Welcome!

We want to use this discussion as a place to share why some of the design and tech choices were made in our project. Feel free to share your opinions and ideas here, but please be mindful of the following:

• Be respectful of everyone's ideas even if they do not sound appealing to you.
• Be as concise as possible in expressing your ideas. Longer discussions will most probably be ignored.
• Share resources and references whenever needed. Cite authors if you are borrowing someone else's idea.
• Enjoy!

[tgarg10]

Why Authenticate with Google (as opposed to creating our own Authenticator)?

Broadly speaking, there are two ways to offer authentication:

• using Google to set up a "Log in with Google" button
• writing our own authenticator that users can login to using their email and password.

Considering the scale of our project, it is highly unlikely that we will be able to support both. While I hate having to be dependent on Google to authenticate our users (adding dependencies increase complexity), I think it saves us a lot of time and code. If we take the path of authenticating our users ourselves, we will have to:

• use a mailing client, which adds more complexity and costs more.
• program and maintain APIs to authenticate users, set up a way to reset passwords (for both "forget passwords" and on-demand), and set up a way to log out.

I think using Google's Gmail authenticator saves us that effort. It does, however, tie us to Google.

[tgarg10]

More thoughts:
Creating our own authenticator would allow us to have the functionality to send emails on demand to the user. For example, we could notify students through email about new assignments or due-dates. Thinking about local authentication 🤔

[datandrews]

Here are some thoughts: - UD is tied to Google as an institution, so it makes sense to tie this app to Google too. This would also fit into UD's single sign on architecture. - If we were thinking of offering something to a broader community --- other institutions, perhaps --- then a non-Google authentication process could make sense, but I'd put this as very, very low priority. - I have used the Google APIs to send automatic emails. We'd probably need an account ***@***.****), but that should be something we can request and will likely get approved. There are limits on the free tier, but it's something on the order of 1000s per day. I really think Google authentication is the way to go. D.A. * I'd feel bad if Arthur Utograder enrolled, but I think he's leaning toward Texas A&M...

…

On Wed, Jan 1, 2025 at 10:56 PM Tanish Garg ***@***.***> wrote: More thoughts: Creating our own authenticator would allow us to have the functionality to send emails on demand to the user. For example, we could notify students through email about new assignments or due-dates. Thinking about local authentication 🤔 — Reply to this email directly, view it on GitHub <#28 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAK635ZIILI4YYVQ5ODFT6T2ITBIRAVCNFSM6AAAAABUPKLNP6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCNZRGM4TMMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[tgarg10]

I actually did read this way back it was published but forgot to reply. I do think that the reasons that @datandrews mentioned make perfect sense. I know our priority is to get the project over the line and bug free before the start of the semester, but this might be a good feature to add in a coming iteration. Maybe as an option at least. @Max-Harold @a2k6 @Charles-Dea