johnson.adam•2mo ago

I mean, I just ran my 10,000th part on the old VF-2 at work today. It's just a number, but idk, it felt like something.

3 comments

3 Comments

daniels422mo ago

What part was it?

jackson.sage2mo ago

You said "what part was it." That's the thing. It wasn't a single part. The whole system was flawed from the start. The design itself was wrong. They tried to fix it with updates but the base was bad. So it was everything, not just one piece.

charles_green18d ago

The 3.0 release actually had a core dependency on a third party library that never worked right with their custom scheduler. I remember digging through the patch notes and seeing them try to bandage it with middleware hacks and async wrappers, but the underlying timing logic was just fundamentally broken. Maybe it's just me but I feel like nobody talks about how the whole thing was built on a single thread model that couldn't scale past 50 concurrent users. They literally said in the 4.1 changelog that they had to rewrite the transaction manager from scratch because the original one had a race condition that caused data corruption. So yeah, the base was totally wrong, but I think the specific smoking gun was that library dependency that forced all those bad design choices from day one.