Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Scaling LLMs to Larger Codebases

blog.kierangill.xyz

Post reply on HN
Print view View original

111–120 of 123 posts

Re: Scaling LLMs to Larger Codebases

#111

Post by KptMarchewa »

the_sleaze_ wrote: post #79
coldtea wrote: post #76

Earlier quoted context omitted.

> Instead of considering that something should be written encapsulated into an object to maintain state, it will instead write 5 functions, passing the state as parameters between each function. Sounds very functional, testable, and clean. Sign me up.

I know this is tongue in cheek, but writing functional code in an object oriented language, or even worse just taking a giant procedural trail of tears and spreading it across a few files like a roomba through a pile of dog doo is ... well.. a code smell at best. I have a user prompt saved called clean code to make a pass through the changes and remove unused, DRY and refactor - literally the high points of uncle bob…

>I know this is tongue in cheek, but writing functional code in an object oriented language, or even worse just taking a giant procedural trail of tears and spreading it across a few files like a roomba through a pile of dog doo is ... well.. a code smell at best.

After forcing myself over years to apply various OOP principles using multiple languages, I believe OOP has truly been the worst thing to happen to me personally as engineer. Now, I believe what you actually see is just an "aesthetics" issue, moreover it's purely learned aesthetics.

Re: Scaling LLMs to Larger Codebases

#112

Post by mstank »

master_crab wrote: post #107
nextaccountic wrote: post #101

Earlier quoted context omitted.

> As of Dec 2025, Sonnet/Opus and GPTCodex are both trained and most good agent tools (ie. opencode, claude-code, codex) have prompts to fire off subagents during an exploration (use the word explore) and you should be able to Research without needing the extra steps of writing plans and resetting context. I'd save that expense unless you need some huge multi-step verifiable plan implemented. Does the UI shows clearl…

The UI (terminal) in Claude code will tell you if it has launched a subagent to research a particular file or problem. But it will not be highlighted for you, simply displayed in its record of prompts and actions.

If you use the vscode extension you can click to view the sub-agent prompts and see all tool calls.

Re: Scaling LLMs to Larger Codebases

#113

Post by saxenaabhi »

prmph wrote: post #109
saxenaabhi wrote: post #104

Earlier quoted context omitted.

fyi that happened to me with codex. but, why is it a big issue? if it does something bad, just reset the worktree and try again with a different model/agent? They are dirt cheap at 20/m and I have 4 subscription(claude, codex, cursor, zed).

The issue is that if it's struggling sometimes with basic instruction following, it's likely to be making insidious mistakes in large complex tasks that you might no have the wherewithal or time to review. The thing about good abstractions is that you should be able to trust in a composable way. The simpler or more low-level the building blocks, the more reliable you should expect them to be. In LLMs you can't really…

I'm not sure you can make that assumption even when a human wrote that code. LLMs are competing with humans not with some abstraction.

> The issue is that if it's struggling sometimes with basic instruction following, it's likely to be making insidious mistakes in large complex tasks that you might no have the wherewithal or time to review.

Yes, that's why we review all code even when written by humans.

Re: Scaling LLMs to Larger Codebases

#114

Post by xnorswap »

nextaccountic wrote: post #101
jarjoura wrote: post #59

Earlier quoted context omitted.

As of Dec 2025, Sonnet/Opus and GPTCodex are both trained and most good agent tools (ie. opencode, claude-code, codex) have prompts to fire off subagents during an exploration (use the word explore) and you should be able to Research without needing the extra steps of writing plans and resetting context. I'd save that expense unless you need some huge multi-step verifiable plan implemented. The biggest gotcha I found…

> As of Dec 2025, Sonnet/Opus and GPTCodex are both trained and most good agent tools (ie. opencode, claude-code, codex) have prompts to fire off subagents during an exploration (use the word explore) and you should be able to Research without needing the extra steps of writing plans and resetting context. I'd save that expense unless you need some huge multi-step verifiable plan implemented. Does the UI shows clearl…

Yes it will, this is almost verbatim (redacted product) claude-code output from my current session:

   ● I'll explore the codebase to understand the current  architecture, testing patterns, and integration points. This will help me formulate effective strategies for reducing QA burden.

   ● 3 Explore agents finished (ctrl+o to expand)
      ├─ Explore  architecture · 57 tool uses · 60.0k tokens
      │  ⎿  Done
      ├─ Explore current testing approach · 29 tool uses · 51.7k tokens
      │  ⎿  Done
      └─ Explore API integration patterns · 44 tool uses · 71.7k tokens
         ⎿  Done
During agent execution, it also shows what each sub-agent is up to. In ctrl+o mode it'll show the prompts it passed to each sub-agent.

Re: Scaling LLMs to Larger Codebases

#115

Post by some-guy »

andrewmutz wrote: post #10

The issues raised in this article are why I think highly-opinionated frameworks will lead to higher developer productivity when using AI assisted coding You may not like all the opinions of the framework, but the LLM knows them and you don’t need to write up any guidelines for it.

I can vouch for this as someone who works in a 1.6 million line codebase, where there are constant deviations and inconsistent patterns. LLMs have been almost completely useless on it other than for small functions or files.

Re: Scaling LLMs to Larger Codebases

#116

Post by dboreham »

jarjoura wrote: post #59
mstank wrote: post #16

As the models have progressively improved (able to handle more complex code bases, longer files, etc) I’ve started using this simple framework on repeat which seems to work pretty well at one shorting complex fixes or new features. [Research] ask the agent to explain current functionality as a way to load the right files into context. [Plan] ask the agent to brainstorm the best practices way to implement a new featur…

As of Dec 2025, Sonnet/Opus and GPTCodex are both trained and most good agent tools (ie. opencode, claude-code, codex) have prompts to fire off subagents during an exploration (use the word explore) and you should be able to Research without needing the extra steps of writing plans and resetting context. I'd save that expense unless you need some huge multi-step verifiable plan implemented. The biggest gotcha I found…

AI can be an FP absolutist too.

Re: Scaling LLMs to Larger Codebases

#117

Post by dboreham »

zingar wrote: post #98
layer8 wrote: post #64

Earlier quoted context omitted.

This is a bit like agile versus waterfall.

100%, the reason I thought of this is constantly telling developers to break their work down into smaller pieces so that they can focus and the customer sees value sooner.

One of the things I like about LLM coding is that I don't need to become a psychologist in order to persuade other humans to approach their work in an manner I'd prefer.

Re: Scaling LLMs to Larger Codebases

#118

Post by justsid »

throw-12-16 wrote: post #92
lnx01 wrote: post #46

LLMs are so good at telling me about things I know little to nothing about, but when when I ask about things I have expert knowledge on they consistently fail, hallucinate, and confidently lie...

You just don't know enough to identify the bullshit when you aren't an expert in that domain.

That’s the joke

Re: Scaling LLMs to Larger Codebases

#120

Post by ewoodrich »

PeterFBell wrote: post #106
ewoodrich wrote: post #82

Earlier quoted context omitted.

I recently ran into two baffling, what felt like GPT 3.5 era completely backwards misinterpretations of an unambiguous sentence once each in Codex and CC/Sonnet a few days apart in completely different scenarios (both very early in the context window). And to be fair, they were notable partially as an "exception that proves the rule" where it was surprising to see but OP's example can definitely still happen in my ex…

It's complicated. Firstly, don't love that this happens. But the fact you're not willing to provide tolerance to a commercial tool that costs maybe a few hundred bucks a month but are willing to do so for a human who probably costs thousands of bucks a month is revealing of a double standard we're all navigating. Its like the fallout when a waymo kills a "beloved neighborhood cat". I'm not against cats, and I'm deepl…

I mean, we typically architect systems depending on humans around an assumption of human fallibility. But when it comes to automation, randomly still doing the exact opposite even if somewhat rare is problematic and limits where and at what scale it can be safely deployed without needing ongoing human supervision.

For a coding tool it’s not as problematic as hopefully you vet the output to some degree but it still means I have don’t feel comfortable using them using them as expansively (like the mythical personal assistant doing my banking and replying to emails, etc) as they might otherwise be used with more predictable failure modes.

I’m perfectly comfortable with Waymo on the other hand, but that would probably change if I knew they were driven by even the newest and fanciest LLMs as [toddler identified | action: avoid toddler] -> turns towards toddler is a fundamentally different sort of problem.

Post reply on HN