Comment by Abdullah Oladipo on "Building a simple "overengineered" project."

Hi Taiwo, thanks for the amazing article, particularly learned something new from the socketIO section, your explanation of how it was work was brilliant.

As for Durator, it didn't work for me, it asked for a link to any media file and I put a youtube link but Generate button wouldn't click; As for metaworka, I uploaded a png thumbnail file but got a response that it was a video and the frame rate was 25 fps.

I also have a question about the diagram under caching for speed section: why is cache a separate block when it is actually been handled by the same Redis, you mentioned that you map the URL to a UUID and to the result, shouldn't it go to Redis straight, get processed by FFPROBE and return back to Redis cos the cache looks like a separate entity

something like this

Thank you for your questions. I will get to the developer of Durator and Metaworka to fix those.

The cache is basically the same redis instance and storage. In order to make the design more linear and straight forward, I chose to use them as separate blocks of the same colour. This is great for explaining that a cache is a storage that is read before any process is started. In a bigger project, you might not have the luxury of having redis serve as both volatile storage and cache. Hence, it's great to point that out from the offset.

With regards to your second question, if we know that the cache and the redis storage are one and the same, what you describe is exactly what was shown in the diagram. Get the URL, check if it exists as a key in the cache. If it does, get the uuid stored as a value and check if it also exists as a key in that same cache. If it does, return the result. If not, queue it and wait for it to complete. Any failure will also queue the request.

I agree that the drawings could be better and I hope it was at least good enough to pass the information through.

Thanks

Thanks Taiwo Hassan Yusuf

I didn't even notice the colour coding, thanks for expanding on that, that's another tip to add to what I learned from this article.

Still wondering why every request passes through the FFMBPROBE when some could already be cached and mapped url-uuid-result

Abdullah Oladipo They don't all do. Some return fast while others go through. I did a poor job of showing it though.

Taiwo Hassan Yusuf

Alright then, awesome 🙌

As for durator, I think it was shipped with the tagline 'any file' wrongly.

Works fine with audio and video files

Abdullah Abdulfatah Is it meant to be hosted somewhere?

Cos I added link to YouTube video but failed to work. Thanks

awesome done Abdullah Abdulfatah

Search Hashnode