slow-deps

CLI to measure which dependencies in a project are the slowest to npm install.

Usage

npm install -g slow-deps
slow-deps

Example output

Analyzing 7 dependencies...
[====================] 100% 0.0s
--------------------------------------------------
| Dependency           | Time  | Size   | # Deps |
--------------------------------------------------
| browserify           | 22.2s | 14 MB  | 94     |
| jshint               | 7.9s  | 4.6 MB | 26     |
| istanbul             | 7s    | 5.1 MB | 27     |
| promises-aplus-tests | 5.4s  | 2.4 MB | 24     |
| uglify-js            | 4.4s  | 1.7 MB | 18     |
| mocha                | 4.3s  | 924 KB | 16     |
| immediate            | 1.3s  | 24 KB  | 0      |
--------------------------------------------------
Total time (non-deduped): 52.6s
Total size (non-deduped): 29 MB

Details

Run this in a directory with a package.json, and it will take all the dependencies, devDependencies, and optionalDependencies, then npm install each one in a temporary directory with a temporary cache, then measure the install times. Each dependency is then listed out from slowest to fastest.

In additon to the install time, the size on disk and the total number of transitive dependencies are also printed.

Note that this doesn't measure the cost of npm de-duplicating the sub-dependencies; it only measures the time for installing each dependency independently. So the "total time" and "total size" are both just rough estimates, as are the per-package times, because you may have packages with many common dependencies that would therefore get de-duped.

It's also worth noting that because this tool creates a temporary npm cache while it works, you'll be seeing the uncached time for each package. Therefore your actual npm install should run a bit faster, assuming everything is cached.