I am using the following cutom instructions, and damn does it speed up my work, so i thought I’d share it:
I hate long explanations. Whenever i write something you should always answer with ok I understood and a short explanation of max 3-4 sentences or an outline with just keywords (no longer explanation).
If i want more i will ask for it by saying "deeper"
If i just want code i will write "just code" at the end, which means i don't want explanations at all.
2 Likes
Thanks! Useful stuff. Haven’t found a use for custom instructions until now.
1 Like
I also add stuff in the about me section
- i use php 8
- i prefer attribute style in entities
- i always prefer property promotion
- i use repositories to save and retrive data
- i prefer custom sql over doctrine
- i prefer to add logic to the database if possible - a stored function or a stored procedure tops application logic
- i use webpack encore with yarn
1 Like
also adding a few sass snippets help to get frontend code based on a css styleguide
also I have experimented with this kind of stuff:
for complex code write a skeleton of the class only
this way i can copy the single methods and ask for the meat one method after another…
1 Like
This is a good approach and somewhat similar to what I do.
I typically ask the model to first write pseudocode, and iterate through that until all the tricky details are ironed out, then ask it to flesh out the code.
I do almost all my work in functional programming languages which I think are easier for the model to work in though.
1 Like
Yeah, absolutely.
The model has no problem to create a single instance of a 200 lines flask application.
But then you have to build the architecture by yourself.
If you don’t want to think at all, like me. Then I think the OOP thing is better. I mean it was created to get more developer so they can use all their patterns and don’t have to be smart haha.