I have an app that I used Replit Auth in and everything was working great up until I made a change to bypass the authentication flow in the development environment. Now in the production environment every time a user hits “Allow” on the Replit permissions page all it seems to do is refresh, and not complete the authentication session.
Has anyone had success by removing Replit Auth from their project before and starting the Authentication process from scratch?
I’ve said before hot swapping auth is really a PITA, and often results in a lot of time, checkpoints and effort in debugging. But, you should do it now as the app only grows in complexity over time. No time is better than right now to start this painful process.
Yea that’s exactly what I’m feeling. I really don’t want to have to do it, but I think the code is already getting out of hand. Thanks for the input, I’ll see what I can get completed tonight.
What auth are you going to implement? I have a video about Firebase auth if you’re interested.
Send it over my man! That would be a great help!
Debating between firebase and Outseta. I used firebase in Replit in the past, but had a headache with getting the Stripe Webhook to work.
I’ve used Outseta for Webflow clients in the past and am considering using their React setup.
lol this… the auth. Section is awful and causes massive delays for sure. I’ve had one successful attempt at building a prototype that lets users sign up and it requires valid emails and passwords. However, trying to implement outside the box authentication methods is AWFUL. However, it has forced me to start brute forcing the code changes myself instead of using the agent or chatbot… just like learning array structures is probably a good idea too… LLMs (imo) struggle with structure.
Books/ topics you may want get technical material for:
JavaScript
HTML
React
Python (I liked Tony Gaddis 5th edition)
The key is planning. Take the app seriously from the start, give Agent the long-term framework from the start. This frontloads a bit for a simple prototype but it pays dividends!
That is a fact. The ones I jump into or “ape” into are ones that fail pretty quickly despite them being a cool idea lol the ones I take very seriously and try to make all of my prompts before then copy paste and work with the agent do better… I like to write all my prompts and then for kicks I’ll ask all the LLMs on the market (that I know and use) something like, “You’ve been asked what could make these prompts more effective for another LLM to code” or something and just paste my prompts and you’d be surprised how they can reshape your wording (especially if overworked and just exhausted) and then slowly for each component for each block of code have yourself some prompts saved so you can just get right back into the same spot. If you find lines of code or sections that give you issues often I highly recommend using comments in the code document and then log it in your notes so you’re not constantly sifting through huge codebases like a dork lol if you are methodical from day one and take it slow it works a lot better. Again, you are right though any piece of software you develop or engineer should be done with careful planning and R&D lol the problems I have is when it becomes so multi faceted and I’m working 80+ hours and you get lost in the “vibes” and start to unifiy with the vibes and lose yourself lol
I have a similar workflow, very structured, that I follow each time I kick off an app.
Yeah if I don’t my mind gets squirrelly (like my posts here) and I’ll be banging my head against a wall lol