Not all questions are gen ai, but I understand your point. Initially I thought that gen ai could be a huge help, then realized the questions weren't good and there were some mistakes. I tried to comb through them, but I'm only human :-(
This is good feedback, thanks! A mode without a timer is totally doable and shouldn't be too difficult to implement -- I'll add to the list.
That particular question on the number of apples also confused the hell out of me. I checked with Gemini and ChatGPT and both agreed on the answer. Not being a native speaker, and because it was suggested by a friend, I thought I should accept it. I'm sorry it's causing confusion.
Hey, thanks! The core is really simple: Django + Postgres, Tailwind CSS for the UI, and little bits of plain JavaScript for things like the countdown timer.
I also used a couple of libraries like django-allauth for account management and django-storages for managing static assets in S3.
The whole thing runs on AWS.
I have no corporate secrets here, so if there's anything else you'd like to know, happy to share!
Quite a few are AI generated and even though I tried to read through all of them once I realized that often the choices weren't great, I probably missed a few :-(
That's a good example, and I'll fix it now. FWIW, I also have it on my list to add a feature to flag bad questions.
Are you referring to the interruptions from one question to the next that show the explanation? Those were added on purpose because originally I had this in mind for kids in the 8 -- 16 years old range and wanted it to be educational.
But I see what you mean, and it's no trouble to add an option for "rapid mode" to just blaze through questions! I'll add this to the list!
There's actually a (not yet documented) API available. If you curl curl https://www.pastesafely.com/api/paste/P73 you'll the paste and it's metadata in json. Would that work for your use case?
Since it is a pastebin, it would be nice to have something that just spit out the raw content of the pasteboard. I guess E2E encryption will not work here but often times people don't need that. Being able to tell my user to just type:
Let me put all suggestions on this together. They should be fairly easy to implement and will enhance the usefulness and security of this feature:
* You should support HTTP HEAD request
* You should infer MIME type from language selections and set Content-Type header as much as possible. If there is none, the default shall be text/plain not text/html
* You should set `Content-Security-Policy: sandbox` HTTP header so people cannot use your service to do malicious stuff
* You should set `Access-Control-Allow-Origin: *` so it can be used in pure client-side JAMStack applications
Thanks. If you can infer the correct MIME type from the language and send the correct Content-Type header it will be even more useful. I know most languages don't have a dedicated MIME, but some do, such as html, json or svg.
