DeepWiki AI analysis

[Liam OBrien]

In the past, I submitted the Ring-related GitHub repos to DeepWiki, an AI analysis tool from Cognition AI. Recently, I resubmitted the following repos so that the analysis could be brought up-to-date:

1. Ring Programming Language (deepwiki.com/ring-lang/ring)
2. Ring Website (deepwiki.com/ring-lang/ring-lang.github.io)
3. PWCT2 (deepwiki.com/PWCT/PWCT2)
4. Softanza (deepwiki.com/mayouni/stzlib)

I found the DeepWiki analysis is comprehensive, but the generated diagrams are not always correct. I would be curious as to how you find the analysis? Does it lean too much towards developers, for example? 

To show how DeepWiki handles the repo content, here is the outline generated for the Ring Programming language:

Ring Programming Language Overview
  What is Ring
  Architecture Overview
Core Language System
  Ring Virtual Machine
    Bytecode Execution and Instruction Set
    Function Call Mechanism
  Scope and Variable Management
Memory Management and Garbage Collection
  Garbage Collection System
  Memory Pool Manager
Data Structures and Collections
  List Implementation
Language Features and Syntax
  Natural Language Programming
  Syntax Flexibility
  Multi-Paradigm Support
Extension System
  C/C++ Extension Architecture
  Code Generator System
  Ring API Functions
Documentation System
  Documentation Build Pipeline
  Search System
  Release Notes and Language Evolution
Application Development
  Web Development
    HTTPLib and WebLib
    Deployment Options
  Package Ecosystem
    RingPM Package Manager
    Available Packages and Extensions
  GUI and Game Development
Development Infrastructure
  Testing System
    Test Organization and Categories
    Test Execution and Validation
  Security and Validation
  Cross-Platform Support
  Build System and Code Generation
Community and Distribution
  Website and Marketing
  Applications Showcase
  External Integrations

As a counterpoint example, here is the summary generated for the Softanza library: 

Summary: The Softanza Philosophy in Practice

Five Principles, One Vision

Softanza's design philosophy creates a cohesive programming experience where:

1. Code reads like thought (Natural Orientation)
2. Strings flow naturally, objects empower (String-First with Q())
3. Errors are prevented, not handled (Safety by Design)
4. Multiple paths serve different minds (Dual APIs and Flexibility)
5. Speed comes from Qt, clarity from Ring (Qt Foundation)

Might DeepWiki be a useful supplement to understand the inner workings of a GitHub repo and in turn help with the development of the Ring-related ecosystem?

Kind Regards,
Liam

[Mahmoud Fayed]

Hello Liam

Thank you very much for sharing such beautiful links.

I enjoyed reading all of the content on this page:  ring-lang/ring | DeepWiki

Very impressive, high‑quality results and analysis. The diagrams are beautiful, and after spending many years working on Ring, seeing diagrams like these truly touches my heart :D 

There’s just a small mistake in the Extension Development Workflow. The code shown in the configuration file is not correct.

In my opinion, the accuracy of this article is 99.99%.

Greetings,

Mahmoud

[Liam OBrien]

You are most welcome, Mahmoud!

By the way, as the repo owner, you can use the "Edit Wiki" button to make changes.

Kind regards, 

Liam

[Mansour Ayouni]

Hello Liam,

The tool is impressive, still it needs further prompting to capture Softanza architecture and features.

While most of the things said are rather correct, they are not presented the accurate way.

The main misconception made is to look at Softanza, architecturally, as an integration system of 7 subsystems.

Also, many foundational features of the library are not covered.

The reason may be the large size of the codebase that exceeds the token window of the tool.

Still, I'm very happy you made me discover the tool, and I'll dedicate some time to better educate it to better document Softanza.

Thank you very much!

Best regards,

Mansour

--

---
You received this message because you are subscribed to the Google Groups "The Ring Programming Language" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ring-lang+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ring-lang/CAL%2BJo2b3XbrJQgXcytkFOhPaZk5%2BD1YC0W%2B3nwsy%2BxdeqeWRXA%40mail.gmail.com.

[Liam OBrien]

Hi Mansour,

Yes, the context window is a critical parameter in interfacing with large language models. But even with a proper context, they may not always assess the content accurately. 

Early in the summer, I indexed a repo for another developer and he mentioned that the results were surprisingly accurate for what was a little-used language. However, example code generated by the LLM would not have been done in real life and was in fact an hallucination. So, even with a smaller amount of content, it can generate inaccurate assessments.

I would be curious if you can overcome the DeepWiki limitations and generate a more accurate picture of your Softanza repo. Do you find that chatting with the repo using DeepWiki gives accurate results?

Kind Regards,
Liam