AI and an HTTP caching bridge

[Yaron 60]

I have a need for a very specific feature I'd like to add to Zoom Player, an HTTP caching bridge.

 

The purpose of an HTTP caching bridge is to cache repeated HTTP "GET" queries generated by DirectShow media streaming filters (components) such as LAV Filters when streaming mp4/mkv files from media servers such as PLEX, Emby or Jellyfin.
 

Caching is required as these components treat streaming files the same as local files with repeated seeking to read headers and frame indexes, degrading performance (very slow seeking, long pauses when switching subtitle tracks, etc) and unnecessarily overloading the media server.

 

Since the HTTP caching bridge is a 100% self-contained feature, I thought AI would be well suited to the task.

 

I wrote a very detailed design document and fed it into every AI system I had access to and asked the AI to implement the project goal (a section in the design document).

 

I wrote about my experiences trying to get various AI models to do the job and included the design document at the end of the post:

https://www.reddit.com/r/ZoomPlayer/comments/1m0fy46/ai_and_an_http_caching_bridge/

 

Has anyone managed to get AI to code anything like this level of complexity?

[Anders Melander 2050]

3 hours ago, Yaron said:

Has anyone managed to get AI to code anything like this level of complexity?

No. You are wasting your time.

[Yaron 60]

Possibly, but soon (12-18 months) one of these AI models will be able to do it close enough that I can make it work and I can wait, I have other more important features in my work-queue.

 

For now this is my personal benchmark for AI's coding capabilities.

[Geoffrey Smith 52]

On 7/16/2025 at 1:47 AM, Yaron said:

I have a need for a very specific feature I'd like to add to Zoom Player, an HTTP caching bridge.

 

I wrote about my experiences trying to get various AI models to do the job and included the design document at the end of the post:

https://www.reddit.com/r/ZoomPlayer/comments/1m0fy46/ai_and_an_http_caching_bridge/

 

Has anyone managed to get AI to code anything like this level of complexity? 

I would split it up into multiple parts. Make the code be generated into multiple separate files (you likely can combine them back later if that is important to you). Focus on getting each part correct.  separate out the different parts... HTTP server part, request decoding,  cache file storage/ range checking,  HTTP request.  I suspect concurrency is going to be a big issue so tell the AI to make sure it takes that into account.

[Yaron 60]

If I'm going to that level of detail in the code itself, I can write it myself.

 

This is a form of benchmark for me, a nice to have feature that isn't killing me right now as there are alternatives (e.g. using libVLC media engine).