Examples are not running
Its incredibly painful to use the system, the examples are not running and it would have been amazing to see it run in the cloud; setting up locally is bad UX
15 Replies
@geniusgeek thanks so much for letting us know - which example did you try to run and couldn't?
I tried the weather app
I get this error: then I try to format it
I’ve noticed that when things break at this point, it’s always hard to fix
Now, I did this after the fix, but from my experience it still fail to work
I get this error:
But still listens on port 3001; and the message there was “hello world”
@matijash please what’s the fix?
@martinsos @geniusgeek is talking about the generated apps
@geniusgeek thanks for reaching out!
We are working on fixing this this week -> GPT is still doing these silly small mistakes, but we have some good ideas how to handle those, so hopefully in a couple of days there will be much less of these (wrong prisma relations, wrong imports).
In the meantime, as for specific import errors that you have, this is what imports should look like:
-
import { useQuery } from '@wasp/queries'
- import { useAction } from '@wasp/actions'
- import createLocation from '@wasp/actions/createLocation'
- import fetchWeather from '@wasp/queries/fetchWeather'
- import getWeather from '@wasp/queries/getWeather'
- import getLocation from '@wasp/queries/getLocation'
I am guessing here from the names what is query and what is action though, but this is probably correct.
I hope this helps! Again, sorry about all the small mistakes -> sometimes GPT does more of these mistakes, sometimes less, but we will likely have somewhat better system in a couple of days!
As for deploying to cloud -> that would be great, but since they are full stack web apps with FE + BE + DB it is also quiet demanding both on the resource and implementation side so it is not something we will be doing at the moment -> maybe in the future though if people will enjoy using it!sounds good: i think people just want to see and play with the outcome with out having to go through the stress of running locally
Wohooo @geniusgeek, you just became a Waspeteer level 1!
people will enjoy it that way
secondly: is it possible for people to connect their existing product to extend features or capabilities? @martinsos ?
Ah yes that would be amazing, it is just super hard to do due to GPT's tendency to hallucinate / make up stuff!
Connecting existing product -> did I understand it correctly. you are talking about providing it with an existing codebase and then having it add functionality? If so then no, right now it is focused on creating initial codebase for a web app in Wasp.
yes, from prototype/MVP to V1
that will be very interesting, if the LLM is trained on my codebase and understands context
i can describe and have it build
@martinsos @matijash : i will like to chat further about this; i believe in automating development.
maybe we could build AWS for Engineering
Do keep in mind that most of the work we did on Magic App Generator is about teacthing ChatGPT via prompts on how to use Wasp and generate an app in it -> we are not using an additionaly trained LLM (we could maybe in the future).
I think there is a lot of potential in that space, and I am sure there will be a lot of developments coming, but our focus at Wasp is very much on the Wasp as a web framework -> that is our central focus, while Magic App Generator is an experiment built on top of Wasp + GPT. That means our resources to work on it are quite limited, as most of our resources are focused on improving and building Wasp web framework itself. So our current goal is to improve the generator to have less mistakes and we probably won't be pushing much more into the LLM direction right after that - possibly in the future if that will be a fit for Wasp.