Sabitlenmiş Tweet
Ever thought: "That's gotta be it. the last idea I will have in this space."..
"So I was thinking..."
English
Bryan MacLee
896 posts
Bryan MacLee
@BryanMaclee
My vocation is, truck driver specializing in the oil and gas industry. My avocation is writing code.
Vernal Ut Katılım Mart 2022
179 Takip Edilen84 Takipçiler
@samlambert I'm sorry you see that as a curse.
That is a curse
English
no matter how late i go to sleep my eyes ping open at 6am. truly cursed.
English
@WarrenInTheBuff loops were pretty self evident, and boring, honestly.
Graphs are a much more interesting conversation.
English
@RyanEls4 @michael_c_law What a positive spin on the dumpster fire that is type coercion
English
@michael_c_law Yes. And if they were both using JavaScript there wouldn't have been any issues 😂
English
( I guess I should have @ThePrimeagen in there somewhere)
... So my graphs are composed of a combonation of md docs whith a slightly enhanced syntax, strict contracts +every part of these "graphs" that can be inferred and checked for staleness, deterministically / programatically, is.
Thats basically it for the rank and file writing TS, React, blah, blah
But for those few brave adventurers putting scrml to the test right now; the "graphs" rabbit hole goes much, much deeper
English
I don't have anything funny or clever to add, so I'll do the lame thing and just give (my) straight answer.
I only know how I use graphs in my workflow.
Once a project gets to a certain point, whatever model you use can't keep the whole thing in context. So every clear session is a blank canvas that has to be re-painted every time you come back to it.
when you reference various parts of the project, agents typically just blind grep the "keywords" in your prompt. Then surface the first correct-on-it's-own answer. This works in the beginning. But, once you have interleaved new features, re-factored, deprecated prototypes and superseded features, etc. What the blind grep yields is often stale.
Personally, I call this the corpus ouroboros effect.
my graphs are a way of allowing an agent to curate the necessary context for any particular thing I'm working on in a session.
Agents can essentially traverse graphs with high-ish confidence in much the same way that humans can quickly understand and interactive flow chart.
English
@techtasium Thank you.
would love to know what you think if you give it a try.
yes I am biased. but, it really is a cool language.
English
Looking to connect people on @X
if you’re into
- vibe coding
- UI/UX
- Claude Code
- Token Maxxing
- build in public
say Hi or drop what you’re working on looking to follow active ones 👊
English
If you could only keep one AI coding tool, which would it be?
• Codex
• Claude
• Cursor
• Gemini
Why?
English
@eiranoirx I am working on a deterministic doc engine, as a companion to the scrml compiler.
Im calling it. D..
English
@ImLunaHey It absolutely was a hell site. But it was dependably and consistently a hell site, at least
English
the best thing about using an LLM over stackoverflow is i dont have to deal with questions/answers being removed for being "offtopic" and the stupid fucking egos of the people on that hell site.
Arvid Kahl@arvidkahl
So we can assume that at this point, every single conversation, question, and reply from Stack Overflow has been ingested into every single frontier AI model, right?
English
Hey
@X
I'm looking to #connect with people interested in:
SAAS
Frontend
Backend
Fullstack
DevOps
Web3
Cloud
Freelance
Open Source
#Letsconnect 🙌✨
English
