Conceptually, the "input" to npm-install(1) is a package.json(5), while its
"output" is a fully-formed node_modules
tree: a representation of the
dependencies you declared. In an ideal world, npm would work like a pure
function: the same package.json
should produce the exact same node_modules
tree, any time. In some cases, this is indeed true. But in many others, npm is
unable to do this. There are multiple reasons for this:
different versions of npm (or other package managers) may have been used to install a package, each using slightly different installation algorithms.
a new version of a direct semver-range package may have been published since the last time your packages were installed, and thus a newer version will be used.
A dependency of one of your dependencies may have published a new version, which will update even if you used pinned dependency specifiers (1.2.3
instead of ^1.2.3
)
The registry you installed from is no longer available, or allows mutation of versions (unlike the primary npm registry), and a different version of a package exists under the same version number now.
As an example, consider package A:
{
"name": "A",
"version": "0.1.0",
"dependencies": {
"B": "<0.1.0"
}
}
package B:
{
"name": "B",
"version": "0.0.1",
"dependencies": {
"C": "<0.1.0"
}
}
and package C:
{
"name": "C",
"version": "0.0.1"
}
If these are the only versions of A, B, and C available in the
registry, then a normal npm install A
will install:
A@0.1.0
`-- B@0.0.1
`-- C@0.0.1
However, if B@0.0.2 is published, then a fresh npm install A
will
install:
A@0.1.0
`-- B@0.0.2
`-- C@0.0.1
assuming the new version did not modify B's dependencies. Of course, the new version of B could include a new version of C and any number of new dependencies. If such changes are undesirable, the author of A could specify a dependency on B@0.0.1. However, if A's author and B's author are not the same person, there's no way for A's author to say that he or she does not want to pull in newly published versions of C when B hasn't changed at all.
To prevent this potential issue, npm uses package-lock.json(5) or, if present, npm-shrinkwrap.json(5). These files are called package locks, or lockfiles.
Whenever you run npm install
, npm generates or updates your package lock,
which will look something like this:
{
"name": "A",
"version": "0.1.0",
...metadata fields...
"dependencies": {
"B": {
"version": "0.0.1",
"resolved": "https://registry.npmjs.org/B/-/B-0.0.1.tgz",
"integrity": "sha512-DeAdb33F+"
"dependencies": {
"C": {
"version": "git://github.com/org/C.git#5c380ae319fc4efe9e7f2d9c78b0faa588fd99b4"
}
}
}
}
}
This file describes an exact, and more importantly reproducible
node_modules
tree. Once it's present, any future installation will base its
work off this file, instead of recalculating dependency versions off
package.json(5).
The presence of a package lock changes the installation behavior such that:
The module tree described by the package lock is reproduced. This means reproducing the structure described in the file, using the specific files referenced in "resolved" if available, falling back to normal package resolution using "version" if one isn't.
The tree is walked and any missing dependencies are installed in the usual fashion.
If preshrinkwrap
, shrinkwrap
or postshrinkwrap
are in the scripts
property of the package.json
, they will be executed in order. preshrinkwrap
and shrinkwrap
are executed before the shrinkwrap, postshrinkwrap
is
executed afterwards. These scripts run for both package-lock.json
and
npm-shrinkwrap.json
. For example to run some postprocessing on the generated
file:
"scripts": {
"postshrinkwrap": "json -I -e \"this.myMetadata = $MY_APP_METADATA\""
}
Using a locked package is no different than using any package without a package
lock: any commands that update node_modules
and/or package.json
's
dependencies will automatically sync the existing lockfile. This includes npm install
, npm rm
, npm update
, etc. To prevent this update from happening,
you can use the --no-save
option to prevent saving altogether, or
--no-shrinkwrap
to allow package.json
to be updated while leaving
package-lock.json
or npm-shrinkwrap.json
intact.
It is highly recommended you commit the generated package lock to source
control: this will allow anyone else on your team, your deployments, your
CI/continuous integration, and anyone else who runs npm install
in your
package source to get the exact same dependency tree that you were developing
on. Additionally, the diffs from these changes are human-readable and will
inform you of any changes npm has made to your node_modules
, so you can notice
if any transitive dependencies were updated, hoisted, etc.
Occasionally, two separate npm install will create package locks that cause
merge conflicts in source control systems. As of npm@5.7.0
, these conflicts
can be resolved by manually fixing any package.json
conflicts, and then
running npm install [--package-lock-only]
again. npm will automatically
resolve any conflicts for you and write a merged package lock that includes all
the dependencies from both branches in a reasonable tree. If
--package-lock-only
is provided, it will do this without also modifying your
local node_modules/
.
To make this process seamless on git, consider installing
npm-merge-driver
, which will teach git how
to do this itself without any user interaction. In short: $ npx npm-merge-driver install -g
will let you do this, and even works with
pre-npm@5.7.0
versions of npm 5, albeit a bit more noisily. Note that if
package.json
itself conflicts, you will have to resolve that by hand and run
npm install
manually, even with the merge driver.