This whole thing is the problem. AngularJS was released in 2010. If in 2010 I'd know that the damn thing would die in 2021, and that I would have to rewrite it all by that date, I would not have used the damn thing in the first place.
I also at some point inherited an app written in Vue 2. By the time I got it, Vue 3 was already out and a couple of years later, Vue 4, completely different to Vue 2, was out. Rewriting was not an option, so I had to create a docker image that can compile the damn thing offline, cause if some part of the supply chain breaks, well, that's it.
Ten or eleven years is not a super long time in enterprise software. Having to keep upgrading and changing libraries just cause the devs of the libraries get bored should not be a thing.
> If in 2010 I'd know that the damn thing would die in 2021
Was it obvious back then? I had to work on an AngularJS site for a while and it was the most confusing thing ever. Having come from React I couldn't understand the complexity but maybe it was still better than callback hell in jQuery?
This is why I struggled and struggled to omit JS at all costs on a greenfield project - many apps from even 5 years ago by now fired teams fail to build with cryptic npm messages. And if you get past that you have to deal with webpack or some such junk. We settled with go fyne which compiles to wasm to handle both local and remotely accessible use cases. Trade off is not as much freedom and a fixed feature set but I know it's gonna work after 5 godamn years.
A problem I found was that getting at a local bus station at London, showed all their destinations as London, but without any precise place of where I would arrive. At one point I traveled from one part of London, to Bash, a completely different city, back to another part of London.
Well yes, but the coincidence that Quarks have charges of multiples of another particle, that is not made up of quarks, should rise your brow, shouldn't it?
Like we could accept coincidences if at the bottom is all turtles, but here we see a stack of turtles and a stack of crocodiles and we are asking why they have similar characteristics even if they are so different.
I've thought about this same idea but it probably gets very complicated.
Let's say, you simulate a long museum hallway with some vases in it. Who holds what? The basic game engine has the geometry, but once the player pushes it and moves it, it needs to inform the engine it did, and then to draw the next frame, read from the engine first, update the position in the video feed, then again feed it back to the engine.
What happens if the state diverges. Who wins? If the AI wins then...why have the engine at all?
It is possible but then who controls physics. The engine? or the AI? The AI could have a different understanding of the details of the base. What happens if the vase has water inside? who simulates that? what happens if the AI decides to break the vase? who simulates the AI.
I don't doubt that some sort of scratchpad to keep track of stuff in game would be useful, but I suspect the researchers are expecting the AI to keep track of everything in its own "head" cause that's the most flexible solution.
Then maybe the engine should be less about really simulating the 3D world and just trying best to preserve consistency, more about providing memory and saving context for consistency than truly simulating a lot besides higher level concerns (at which point we might wonder if it couldn't be directly part of the model somehow), but writing those lines I realize there would probably still be many edge cases exactly like what you are describing...
Kiro is such a disaster. It starts well with all the planning, but I haven't been able to control it. It changes files on a whim and changes opinion from paragraph to paragraph.
Also it uses the Claude models but afaik it is constantly changing which one is using depending on the perceived difficulty.
> it uses the Claude models but afaik it is constantly changing which one is using depending on the perceived difficulty
Claude Code does the same. You can disable it in Kiro by specifically setting the model to what you want rather than “auto” using /model.
Tbh I’ve found Kiro to be much better than Claude Code. The actual quality of results seems about the same, but I’ve had multiple instances where Claude Code get stuck because of errors making tool calls whereas Kiro just works. Personally I also just prefer the simplicity of Kiro’s UX over CC’s relative “flashy” TUI.
There were literally boots on the ground. They left, because that's what happens when an invasion ends. It doesn't mean the invasion retroactively never happened.
At this point I believe the anecdotes more than benchmarks, cause I know the LLM devs train the damn things on the benchmarks.
A benchmark? probably was gamed. A guy made an app to right click and convert an image? prolly true, have to assume it may have a lot of issues but prima facie I just make a mental note that this is possible now.
reply