I've been thinking about this topic a lot lately and I find myself in agreement with you vsareto. I'm largely a JS dev by day (working on backend services) but I'm very passionate about low-level programming. I tend to hack around in Rust or C, even starting to hack with assembly language. Heck, I'm even reading the AMD x86_64 architecture books for fun :P.
There are so many challenges trying to understand what your Javascript is doing under the hood. Take Node.js development for example. Your code is running on V8, likely running within a Docker container, perhaps even on a Kubernetes node or other distributed cluster. To truly understand what your code is doing at the CPU level is a _very_ difficult thing to do. More often than not, the performance issues I've run into have to do with topics like: slow performing network communication, poorly thought out algorithm design (excessive code), "noisy neighbors" where other containers are causing performance problems unrelated to your own code (for example, one issue we had with excessive times for DNS resolution was caused by the networking stack/tech in our k8s cluster being misconfigured and doing a GC every 60 seconds for 4-5 seconds at a time), or poor database-access patterns that cause excessive times for queries.
The JavaScript code itself and inefficiencies of what the resulting machine code is seems to rarely be the case for me. But that may also be due to the domain that I work in (it's actually one of the things that has me super bored with the software I work on these days).
To echo your own points, I find myself most often coaching junior developers on data-access patterns and design, minimizing network activity (if you think missing the cache and going to memory is slow for software, just imagine hitting the network and going to different machines), etc.
Like I said though, it may just be due to the domains I work in and being on the backend.
And all of this from a person who is _loving_ learning more about the internals and playing around with lower-level programming. I feel like the more I have learned at that level has made me a much better programmer as well. I wish I would have learned this stuff 10-15 years ago. So I'm torn.
There are so many challenges trying to understand what your Javascript is doing under the hood. Take Node.js development for example. Your code is running on V8, likely running within a Docker container, perhaps even on a Kubernetes node or other distributed cluster. To truly understand what your code is doing at the CPU level is a _very_ difficult thing to do. More often than not, the performance issues I've run into have to do with topics like: slow performing network communication, poorly thought out algorithm design (excessive code), "noisy neighbors" where other containers are causing performance problems unrelated to your own code (for example, one issue we had with excessive times for DNS resolution was caused by the networking stack/tech in our k8s cluster being misconfigured and doing a GC every 60 seconds for 4-5 seconds at a time), or poor database-access patterns that cause excessive times for queries.
The JavaScript code itself and inefficiencies of what the resulting machine code is seems to rarely be the case for me. But that may also be due to the domain that I work in (it's actually one of the things that has me super bored with the software I work on these days).
To echo your own points, I find myself most often coaching junior developers on data-access patterns and design, minimizing network activity (if you think missing the cache and going to memory is slow for software, just imagine hitting the network and going to different machines), etc.
Like I said though, it may just be due to the domains I work in and being on the backend.
And all of this from a person who is _loving_ learning more about the internals and playing around with lower-level programming. I feel like the more I have learned at that level has made me a much better programmer as well. I wish I would have learned this stuff 10-15 years ago. So I'm torn.