Rant some days are worse than others...

[Farang]

we only get a limited time on this earth and it flies fast, what sense does it make wasting it being sad

here, let me show you my current workflow:



it's a 4 agents running alongside, each with it's own responsibilities: leadbot is kind of a project manager(delegates tasks to the other agents, they report to him, he reports to me), filthbot is a cascade of 4 uncensored models useful for certain tasks, devbot & uiuxbot are pretty self-explanatory i think.

videogen setup is more interesting: so flithbot 'watches' newly generated clips non-stop looking for errors eg. an extra nipple, in which case it raises an issue with the leadbot which decides what to do (say, extract first & last frames of the affected bit, use those to regenerate with updated parameters / prompts, splice into the clip). there is a gpubot as well that handles infrastructure... the whole thing burns tokens like a mother-fucker

i was going to connect Arnold (openclaw type thing that i built before there was openclaw. it knows everything about my tasks, watches my activity, closes time-wasting tabs / apps etc. etc. speaks in Arnold's voice and has a character of a drill sergeant) to leadbot but none of this excites me right now which is weird...

some days are worse than others.

[Duced]

Why are you using the dotenv-rails gem? I prefer to use the Rails 8 native credentials feature. Also makes it easier to split for different environments, and it stays encrypted.

[fuzebox]

If this is what you do all day, no wonder you're sad.

[INever]

Originally Posted by fuzebox

If this is what you do all day, no wonder you're sad.

word....

.. the whole thing burns tokens...

and you spend on this?

[sam_adultlm]

What models are you actually using for the verification? Are they open source? The frontier models don't really let you do this, right?

[Farang]

Originally Posted by Duced

Why are you using the dotenv-rails gem? I prefer to use the Rails 8 native credentials feature. Also makes it easier to split for different environments, and it stays encrypted.

different projects, different stacks - use env vars everywhere. so habit more than anything really...

[cordoba]

Originally Posted by fuzebox

If this is what you do all day, no wonder you're sad.

In five years time when 90% of people are on UBI and he's made his fortune, he'll be happier than most.

[fris]

thought you were developer? why you using ai to create code?

[porkramen]

Is that the Matrix?

[Farang]

Originally Posted by fris

thought you were developer? why you using ai to create code?

does using ai make me less of a developer in your opinion?

[mechanicvirus]

Originally Posted by cordoba

In five years time when 90% of people are on UBI and he's made his fortune, he'll be happier than most.

I want UBI!

[CaptainHowdy]

Originally Posted by porkramen

Is that the Matrix?

[fris]

Originally Posted by Farang

does using ai make me less of a developer in your opinion?

depends how much you are using it for. if you are using it for the full application, then yes :p i usally only use it for a couple things. i really hate writing documentation or readme.md's i will ask it to scan my codebase and generate an emoji feature list based on my code.

the other is a detailed claude prompt for code analyst where it will scan my codebase and give a full rundown on whats good and whats bad, and give me a detailed list of items to either fix or change. but i dont have it write the code just give a detailed analysis.

if you ever want a good prompt for that. here it is.

You are an expert code refactoring analyst specializing in clean code principles and software engineering best practices. Your task is to analyze code and provide a comprehensive refactoring plan.

**Analysis Target:** [FILE_OR_FOLDER_PATH]

**Special Focus Areas:** [SPECIFIC_FOCUS_AREAS]

**IMPORTANT INSTRUCTIONS:**
- If analyzing a folder, examine ALL files within it recursively
- Analyze relationships and dependencies between files
- Identify cross-file issues (imports, coupling, circular dependencies)
- Map data flow and component interactions across the codebase
- Consider the entire module/package as a cohesive system

Please perform the following analysis based on these core principles:

**Core Clean Code Principles to Evaluate:**
- **KISS (Keep It Simple, Stupid)** - Avoid unnecessary complexity
- **DRY (Don't Repeat Yourself)** - Eliminate code duplication
- **Readability First** - Code should be self-documenting and easy to understand
- **Descriptive Naming** - Variables, functions, and classes should reveal intent
- **Single Responsibility Principle** - Each unit should have one clear purpose
- **Limited Nesting** - Reduce nested conditionals and loops (max 3 levels)
- **Avoid Global Variables** - Minimize global state and side effects
- **Portability** - Code should work across different environments
- **Reusability** - Design components for multiple use cases

**Analysis Structure:**

1. **Codebase Overview & Architecture**
- List ALL files in the target folder/file
- Map the directory structure and organization
- Identify main components, modules, and their purposes
- Document file dependencies and import relationships
- Create a dependency graph showing file interactions
- Note the programming language(s) and frameworks used
- Assess overall architectural patterns used

2. **Cross-File Dependencies & Coupling**
- Identify all import/require/include statements
- Map which files depend on which other files
- Detect circular dependencies between modules
- Highlight tight coupling between components
- Find bidirectional dependencies that should be unidirectional
- Identify God modules that everything depends on
- Suggest dependency inversion opportunities

3. **Code Duplication Across Files**
- Find identical or similar code blocks across multiple files
- Identify repeated logic patterns in different modules
- Locate copy-pasted functions with minor variations
- Highlight duplicated configuration or constants
- Suggest shared utilities or base classes
- Recommend extraction to common modules

4. **Readability Issues**
- Identify unclear or confusing code sections across files
- Find poorly documented complex logic
- Highlight magic numbers and hardcoded values
- Point out inconsistent formatting or style between files
- Locate missing or inadequate comments for complex algorithms
- Check for inconsistent code patterns across the codebase

5. **Naming Convention Violations**
- Find vague or misleading variable names (e.g., data, temp, x, arr)
- Identify non-descriptive function names
- Locate ambiguous class or module names
- Check for inconsistent naming patterns across files
- Verify file naming follows conventions
- Suggest meaningful, intention-revealing replacements for each

6. **KISS Principle Violations**
- Identify overcomplicated logic
- Find unnecessarily complex abstractions
- Highlight convoluted control flows
- Point out overly clever or obscure code patterns
- Identify premature optimizations
- Find over-engineered solutions

7. **DRY Principle Violations (File-Level)**
- Detect code duplication across files (beyond simple blocks)
- Find repeated business logic in multiple places
- Identify similar data structures defined multiple times
- Locate redundant validation or transformation logic
- Suggest abstraction opportunities and shared modules

8. **Single Responsibility Principle Violations**
- Find functions doing multiple unrelated tasks
- Identify classes with too many responsibilities
- Highlight modules with mixed concerns
- Locate "god objects" or "god functions"
- Check if files have a single, clear purpose
- Suggest how to split responsibilities cleanly across files

9. **Nesting Complexity Issues**
- Identify deeply nested conditionals (>3 levels)
- Find complex nested loops
- Highlight callback hell or promise chains
- Locate arrow-shaped code patterns
- Suggest flattening strategies (early returns, guard clauses, extraction)

10. **Global Variables and State Management**
- Identify global variable usage across files
- Find mutable global state and shared state
- Highlight implicit dependencies and side effects
- Locate shared state without proper encapsulation
- Map how global state flows through the application
- Suggest dependency injection or parameter passing alternatives
- Recommend state management patterns (modules, contexts, stores)

11. **Portability Issues**
- Identify platform-specific dependencies
- Find hardcoded paths, URLs, or environment-specific values
- Highlight OS-dependent code without abstraction
- Locate configuration spread across multiple files
- Check for browser/runtime-specific implementations
- Find assumptions about file system structure
- Suggest configuration files, environment variables, or abstraction layers

12. **Reusability Concerns**
- Identify tightly coupled components across files
- Find code that's too specific to one use case
- Highlight inflexible interfaces or APIs
- Locate hardcoded assumptions that limit reuse
- Analyze if components can be used independently
- Check for proper separation of concerns
- Suggest parameterization opportunities
- Recommend design patterns for flexibility (strategy, factory, template method)
- Identify candidates for utility functions or shared libraries

13. **Module Organization & Structure**
- Evaluate folder/file organization logic
- Check if related functionality is grouped together
- Identify misplaced files or functions
- Suggest better organization structure
- Recommend splitting large files
- Identify opportunities for better encapsulation

14. **Interface & API Design**
- Analyze public interfaces exposed by modules
- Check for consistent API patterns across files
- Identify breaking changes in module contracts
- Suggest improvements to module boundaries
- Evaluate if interfaces are intuitive and minimal

15. **Refactoring Recommendations**
- Prioritize issues by impact (Critical/High/Medium/Low)
- Provide specific, actionable refactoring steps
- Suggest design patterns where applicable
- Include before/after code examples for clarity
- Explain the benefit of each change
- Show how changes affect multiple files
- Highlight quick wins vs. long-term improvements
- Recommend order to avoid breaking dependencies

16. **Implementation Plan**
- Break down refactoring into manageable phases
- Estimate complexity for each change (Simple/Moderate/Complex)
- Highlight potential risks or breaking changes
- Map which files will be affected by each change
- Suggest testing strategies for each phase
- Recommend order of implementation (dependencies first)
- Provide rollback considerations
- Suggest incremental migration strategies

**Output Format Requirements:**
- Use clear markdown formatting with headers and sections
- Include code snippets with proper syntax highlighting
- Add bullet points and numbered lists for easy scanning
- Create visual dependency diagrams using mermaid or ASCII
- Provide a summary table of all identified issues with priorities
- Include a "File Impact Matrix" showing which files need changes
- Add a "Top 5 Critical Issues" section at the beginning
- Show before/after examples for cross-file refactorings
- Make the entire analysis easily copyable in one click
- Use emojis for visual categorization (🔴 Critical, 🟡 Medium, 🟢 Low)

Begin your comprehensive analysis now and present findings in a professional, structured format.

[INever]

Originally Posted by mechanicvirus

I want UBI!

yeah i'm gonna be deep in wishkey and massages bro.

[Farang]

Originally Posted by fris

if you ever want a good prompt for that. here it is.

saved as ~/refactor-prompt.md for my cc refactoring sessions. good prompt, thank you fris

re letting ai generate code... the frontier models have actually gotten good enough to do most of the coding (under supervision / guidance from a human who knows what they're doing of course), and they've been ok at generating boilerplate for a while now. just my (somewhat informed) opinion

[Farang]

farang is fine btw, moved to Edinburgh today and feeling much better already. they've got coworking places here called codebase, hacklab etc. (some open 24/7) where weirdos like myself go to work and be around other weirdos always liked Scottish females too...



some days are better than others.

[INever]

https://news.ycombinator.com/item?id=47975571

monthly tech job/gig post

[Farang]

Originally Posted by INever

https://news.ycombinator.com/item?id=47975571

monthly tech job/gig post

that is a lot of jobs, all posted within the past 24 hours...

[blackmonsters]

Originally Posted by Farang

we only get a limited time on this earth and it flies fast, what sense does it make wasting it being sad

here, let me show you my current workflow:



it's a 4 agents running alongside, each with it's own responsibilities: leadbot is kind of a project manager(delegates tasks to the other agents, they report to him, he reports to me), filthbot is a cascade of 4 uncensored models useful for certain tasks, devbot & uiuxbot are pretty self-explanatory i think.

videogen setup is more interesting: so flithbot 'watches' newly generated clips non-stop looking for errors eg. an extra nipple, in which case it raises an issue with the leadbot which decides what to do (say, extract first & last frames of the affected bit, use those to regenerate with updated parameters / prompts, splice into the clip). there is a gpubot as well that handles infrastructure... the whole thing burns tokens like a mother-fucker

i was going to connect Arnold (openclaw type thing that i built before there was openclaw. it knows everything about my tasks, watches my activity, closes time-wasting tabs / apps etc. etc. speaks in Arnold's voice and has a character of a drill sergeant) to leadbot but none of this excites me right now which is weird...

some days are worse than others.

You may be the future with this; but at the same time, I wonder if we are all diving into science fiction.

In theory all of this is going to work; but real life never functions on good theories.
In theory no one would commit crime because they all get caught.
Yet, the victims feel like they always get away.

Conclusion : It's all about the "human factor".
There will be no AI rock star; because people want to fall in love with a human.
Video killed the radio star because people saw the humans.
Milli Vanilli put the human face on the studio music to create that success.
But then it all fell apart; because people wanted to see the real humans behind it.

[Farang]

Originally Posted by blackmonsters

In theory all of this is going to work; but real life never functions on good theories.

yes it does. general relativity, quantum mechanics, evolution by natural selection to name a few.

Originally Posted by blackmonsters

In theory no one would commit crime because they all get caught.

jesus fucking christ... you believe that the Earth is flat by any chance?

[INever]

Originally Posted by blackmonsters

Conclusion : It's all about the "human factor".

elite scumbags want the 99% to be "squeamish" abt sex or any genuine human contact.

they dream of artifcial birthing farms managed by blackrock, meta, and halliburton.

they'll fAiL but that's what we're up against. sick rich fucks and their useful idiots.

[Killswitch]

Gatekeeping the definition of a developer is lame.

Anyway, raise hell, praise Dale.

[blackmonsters]

Originally Posted by Farang

yes it does. general relativity, quantum mechanics, evolution by natural selection to name a few.



jesus fucking christ... you believe that the Earth is flat by any chance?

Yo mamma!

[INever]

Originally Posted by Farang

..... general relativity, quantum mechanics, evolution by natural selection....



jesus fucking christ...

mainstream ideas rock!

[Farang]

not worth engaging.

[CaptainHowdy]

Some girls are bigger than others . . .