what is the metric to consider a JS app as "VERY large" as the author puts it? If SLOC is used as a proxy then how code from libraries (i.e. code that the team didn't write) are considered? Recently when I generated a Ruby-on-Rails app with webpack enabled, it downloaded 8148 JS files in the node_modules folder. 675K SLOC. I thought it was absolute madness.
The size of node_modules probably isn't a great metric to use if it includes development tools and libraries (and all their transitive dependencies) as well as any libraries you're bundling into the files your user will download and run.
Well, I still have to have those on my computer and in my Docker container. And from a trust, auditing and security perspective it’s pretty horrible, even if multi-stage builds can help
With Buble, Flow, and Roll-up, I can get a decent modern environment with as few dependencies as possible. Upsides? Much faster, too. Less disk space. Docker containers build faster, which is better for CI/CD. And it’s easier to understand the codebase and dependency tree, including development tools.
Just to clarify, I'm not disagreeing with you that the current dependency-heavy JS world is madness or that the bloat during development activities is unhelpful. I'm just saying that having a large/complicated development environment is probably a separate issue to having a large application for most other practical purposes, but looking at the size of node_modules will often conflate the two.
Wow, this must be pretty common. I compiled Bootstrap recently and it download 250 MiB worth of node modules which consisted of 969 dependencies, and 11673 JS files.
