To whoever down-voted me, please do explain how HMRs are comparable to Erlang's true hot code swapping or even Lisp's live redefinition.
HMR is limited to modules (almost or always UI components), and there is no native VM support (module boundaries, side effects, and global state complicates it further) for it, and there is no multi-version coexistence either, and it is absolutely not suitable for production.
To call "hot module replacement" hot loading is very generous, and quite an overstatement.
It is only very superficially similar to hot code swapping. It is a developer tool, not a foundation for live, fault-tolerant systems.
It is absurd to call hot module replacement (HMR) "hot reloading". It might sound fancier, but it is nowhere close to true hot code swapping.
Peak redefinition of terminology. Call it what it is: hot module replacement, at best.
Hot module reloading is common in JS land and does things like trying to preserve running state as opposed to just watching a dir and restarting everything.
HMR is limited to modules (almost or always UI components), and there is no native VM support (module boundaries, side effects, and global state complicates it further) for it, and there is no multi-version coexistence either, and it is absolutely not suitable for production.
To call "hot module replacement" hot loading is very generous, and quite an overstatement.
It is only very superficially similar to hot code swapping. It is a developer tool, not a foundation for live, fault-tolerant systems.
It is absurd to call hot module replacement (HMR) "hot reloading". It might sound fancier, but it is nowhere close to true hot code swapping.
Peak redefinition of terminology. Call it what it is: hot module replacement, at best.