Fixes #25998. Fixes https://github.com/denoland/deno/issues/25928.
Originally I was just going to make this an error message instead of a
panic, but once I got to a minimal repro I felt that this really should
work.
The panic occurs when you have `nodeModulesDir: manual` (or a
package.json present), and you have an npm package with a tag in your
deno.json (see the spec test that illustrates this).
This code path only actually executes when trying to choose an
appropriate package version from `node_modules/.deno`, so we should be
able to fix it by storing some extra data at install time.
The fix proposed here is to repurpose the `.initialized` file that we
store in `node_modules` to store the tags associated with a package.
Basically, if you have a version requirement with a tag (e.g.
`npm:chalk@latest`), when we set up the node_modules folder for that
package, we store the tag (`latest`) in `.initialized`. Then, when doing
BYONM resolution, if we have a version requirement with a tag, we read
that file and check if the tag is present.
The downside is that we do more work when setting up `node_modules`. We
_could_ do this only when BYONM is enabled, but that would have the
downside of needing to re-run `deno install` when you switch from auto
-> manual, though maybe that's not a big deal.
Fixes #25861.
Previously we were attempting to match the version requirement against
the version already present in `node_modules` root, and if they didn't
match we would create a node_modules dir in the workspace member's
directory with the dependency.
Aside from the fact that this caused the panic, on second thought it
just doesn't make sense in general. We shouldn't be semver matching, as
resolution has already occurred and decided what package versions are
required. Instead, we can just compare the versions directly.
`esbuild` can work fine without needing to run post-install script, so
to make it easier on users (especially people using Vite) we are not prompting to run with
`--allow-scripts` again.
We only do that for version >= 0.18.0 to be sure.
Currently we only warn once. With this PR, we continue to warn about
not-run scripts on explicit `deno install` (or cache). For `run` (or
other subcommands) we only warn the once, as we do currently.
Fixes https://github.com/denoland/deno/issues/25862.
npm only makes bin entries executable if they get linked into `.bin`, as
we did before this PR. So this PR actually deviates from npm, because
it's the only reasonable way to fix this that I can think of.
---
The reason this was broken in moment is the following:
Moment has dependencies on two typescript versions: 1.8 and 3.1
If you have two packages with conflicting bin entries (i.e. two
typescript versions which both have a bin entry `tsc`), in npm it is
non-deterministic and undefined which one will end up in `.bin`.
npm, due to implementation differences, chooses to put typescript 1.8
into the `.bin` directory, and so `node_modules/typescript/bin/tsc` ends
up getting marked executable. We, however, choose typescript 3.2, and so
we end up making `node_modules/typescript3/bin/tsc` executable.
As part of its tests, moment executes `node_modules/typescript/bin/tsc`.
Because we didn't make it executable, this fails.
Since the conflict resolution is undefined in npm, instead of trying to
match it, I think it makes more sense to just make bin entries
executable even if they aren't chosen in the case of a conflict.
Refactors the lifecycle scripts code to extract out the common
functionality and then uses that to provide a warning in the global
resolver.
While ideally we would still support them with the global cache, for now
a warning is at least better than the status quo (where people are
unaware why their packages aren't working).
This was initially added in #25399 in order to make transitioning over
from package.json to deno.json more easy, but it causes some problems
that are shown in the issue and it also means that the output of `deno
install` would have different resolution than `npm install`. Overall, I
think it's too much complexity to be smarter about this and it's
probably best to not do it. If someone needs an aliased folder then they
should keep using a package.json
Closes #25538
This allows using npm deps of jsr deps without having to add them to the
root package.json.
Works by taking the package requirement and scanning the
`node_modules/.deno` directory for the best matching package, so it
relies on deno's node_modules structure.
Additionally to make the transition from package.json to deno.json
easier, Deno now:
1. Installs npm deps in a deno.json at the same time as installing npm
deps from a package.json.
2. Uses the alias in the import map for `node_modules/<alias>` for
better package.json compatiblity.
Stores normalized version constraints in the lockfile, which will
improve reproducibility and will fix a bug with duplicate specifiers
ending up in the lockfile. Also, gets rid of some duplicate data in the
specifiers area of the lockfile.
Two small changes:
- In our BYONM errors, suggest running `deno install` instead of `npm
install` if `DENO_FUTURE` is set
- Only emit warning about `deno install` changes if you do `deno install
<foo>` with deno_future unset
This is in preparation for extracting out node resolution code from
ext/node (which is something I'm going to do gradually over time).
Uses https://github.com/denoland/deno_package_json
Regression from
04f9db5b22
Originally I thought to fix the issue in the PR we needed to explicitly
pass through the `node-modules-dir` flag, but after applying the correct
fix that david pointed out (setting `NPM_PROCESS_STATE`) that wasn't
necessary (or correct).
We had a test for deno task with BYONM, but it only tested with
`"unstable": ["byonm"]` in deno.json, so it didn't catch this.
Originally landed in
f6fd6619e7.
Reverted in https://github.com/denoland/deno/pull/24574.
This reland contains a fix that sends "Accept: */*" header for calls made
from "FileFetcher". Absence of this header made downloading source code
from JSR broken. This is tested by ensuring this header is present in the
test server that servers JSR packages.
---------
Co-authored-by: Sean McArthur <sean@seanmonstar.com>
This commit re-implements `ext/fetch` and all dependent crates
using `hyper` and `hyper-util`, instead of `reqwest`.
The reasoning is that we want to have greater control and access
to low level `hyper` APIs when implementing `fetch` API as well
as `node:http` module.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
NPM inserts a default install script when a package has a `binding.gyp`
file.
It's possible, however, for the package to exclude the `binding.gyp`
file when they publish, and in this case the install script will never
succeed for a user of the package.
This happens with `fsevents`, for instance. They don't include the
`binding.gyp` file in their published tarball, but the default install
script appears in the manifest served by `npm`.
This causes us to warn that `fsevents` has an install script, but when
you try to run it it fails due to `binding.gyp` not existing.
Previously when we printed out the packages that skipped install
scripts, we didn't prefix them with `npm:`. When you pass
`--allow-scripts` though, we require `npm:`, which means you can't just
copy paste the package name from the warning message.
