π Like a plumber would say, "Let's-a-go" with part 5
This time, I'll focus a bit more on the development side cause that may be why you've come here in the first place π
1/13
#chatgpt's been great to pop up new projects (albeit some outdated methods once in a while)
But now that we're getting to the meat of software making, which is actually more than following a gettin' started page, a couple of things are happening
Time for numbering #οΈβ£
2/13
1οΈβ£ I need to provide more context more often
Yep, the large data set it's been trained on is not enough. Now it needs to be aware of local context which means:
πΈ Reminding it of folder structure
πΈ Giving files content
πΈ Giving dependencies content
πΈ Fetching it doc
3/13
πΆπ» Let's add two actions
- Copy directory as prompt
- Copy file content as prompt
π€ Erm, I'll put it there, but I'm unsure
πΆπ» Here's the git list-files so you can fix this. Better be the last time I do it manually!
π΄πΎ Narator: "It was"
4/13
Now those prompts are saving me hours π
But there is one issue with gpt4 web
it
it is
it is so
it is so damn
it is so damn slow
π₯± Input is ~100 lines. Response is well above that. So it takes minutes before getting my modified file
π€ May need to change my approach
5/13
2οΈβ£ One function = one file
Yep, the fastest way to remain productive while still using #gpt4. I'm now doing something I loathe: splitting my code into as many small files as I can.
But it allowed me to implement an efficient "CODE_REVIEW" action π¬
6/13
CODE_REVIEW is simple. It's like feeding it the content of a file, but it has to respond back with a fix for every "FIX_ME" comment
Works wonder with #gpt4, but #gpt3.5 was left behind at this point
PSA: I know of other models; I'm forcing myself to use #gpt for this
7/13
3οΈβ£ Fetching dependency context is a drag
#gpt4 works wonders when provided with Typescript typings, but have you looked at those .d.ts files
Well, let me tell you, the webview didn't like it π«
8/13
So I had to copy-paste the entire .d.ts, filter out the declarations manually, add a π
πΆπ» Here are the typings for "openai"
π€ Got it
πΆπ» CODE_REVIEW
π€ Here ya go
πΆπ» π
If you know me by now, yes, this feels like a new tool I'll need π
9/13
But we'll get to that later π°
Cause I'm thinking of creating a blog. Gotta spruce up a place for Twitter haters, right?
Let's take a turn and work on the Chrome Extension cause I believe I'll be getting it out way before the VSCode one
I'm not waiting for the web to get promptful so I'm taking maters into my own hands
1/11
As always, #gpt4 is setting up and creating everything for me admirably. Although I see some questionable configs in eslint, prettier and tsconfig if it tries to provide them
Between pricing (I'm cheap as fπ΅k) and rate limiting, I'm now spending half my time writing
Truth is, I quite enjoy it so let's gooooo π
1/15
So at least the #GPT4's API is available for prompting.
But I don't like the playground
πΆπ» I need pairprog-webview, nextjs, tailwind to chat with your own API
π€ Sure thing
π΅π» Wow, this API call looks quite outdated and doesn't work
π€ ermmm...
In which I make the mistake of continuing my part 1 thread and quickly diverge into some insights about #gpt4 prompting, and the parallels to standards in #leanengineering