Next being owned by Vercel has a perverse incentice to push for features that are good for Vercel, like fast build times and very specific caching strategies, not for the features that developers actually want.
Same. I worked with a Vue 2 project that had cold start time of almost 2 minutes (and ~25s hot start after enabling hardSource plugin in Webpack), and each hot-module reload would take 10+ seconds. It was a real pain to work with. Having build times that are on the order of magnitude faster can really improve the developer experience and the way you work with code.
At this point, vite is pretty damned good and doesn't warrant a replacement.
From the article, the 10x improvement was more 6x and required 30k modules... Turbopack's 15ms was rounded down to 0.01s while Vite's 87ms was rounded up to 0.09s. isn't Android build level slow.
There's always room for performance. I want to use the full capacity of the hardware I bought, why would I intentionally use a slower tool, however slight the difference (all else being equal)?
Aside, this kind of "good enough" attitude is what brought us the JS bundle behemoths, and abstractions on top of abstractions. Even if the difference is small, it adds up if everyone thinks that the performance of their one little library is just "good enough."
I just can't believe they went with the generically named "app" folder...
"pages" is such an obvious name indicating webpages go here. The new "app" directory is super ambiguous, might as well have named it the "code" folder...
Like most of the nextjs 13 changes seem like a lot of extra work for not much benefit.