This commit adds basic support for "node:http2" module. Not
all APIs have been yet implemented, but this change already
allows to use this module for some basic functions.
The "grpc" package is still not working, but it's a good stepping
stone.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
## WHY
ref: https://github.com/denoland/deno/issues/19165
The FileHandle class has many missing methods compared to node.
Add these.
## WHAT
- Add close method
---------
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
Related issue: https://github.com/denoland/deno/issues/19358.
This is a regression that seems to have been introduced in
https://github.com/denoland/deno/pull/18905. It looks to have been a
performance optimization.
The issue is probably easiest described with some code:
```ts
const target = new EventTarget();
const event = new Event("foo");
target.addEventListener("foo", () => {
console.log('base');
target.addEventListener("foo", () => {
console.log('nested');
});
});
target.dispatchEvent(event);
```
Essentially, the second event listener is being attached while the `foo`
event is still being dispatched. It should then not fire that second
event listener, but Deno currently does.
<!--
Before submitting a PR, please read https://deno.com/manual/contributing
1. Give the PR a descriptive title.
Examples of good title:
- fix(std/http): Fix race condition in server
- docs(console): Update docstrings
- feat(doc): Handle nested reexports
Examples of bad title:
- fix #7123
- update docs
- fix bugs
2. Ensure there is a related issue and it is referenced in the PR text.
3. Ensure there are tests that cover the changes.
4. Ensure `cargo test` passes.
5. Ensure `./tools/format.js` passes without changing files.
6. Ensure `./tools/lint.js` passes.
7. Open as a draft PR if your work is still in progress. The CI won't
run
all steps, but you can add '[ci]' to a commit message to force it to.
8. If you would like to run the benchmarks on the CI, add the 'ci-bench'
label.
-->
Internally, `node-tap` spawns a child process with `stdio: [0, 1, 2]`.
Whilst we don't support passing fd numbers as an argument so far, it
turns out that `[0, 1, 2]` is equivalent to `"inherit"` which we already
support. See: https://nodejs.org/api/child_process.html#optionsstdio
Mapping it to `"inherit"` is fine for us and gets us one step closer in
getting `node-tap` working. I'm now at the stage where already the
coverage table is shown 🎉
<!--
Before submitting a PR, please read https://deno.com/manual/contributing
1. Give the PR a descriptive title.
Examples of good title:
- fix(std/http): Fix race condition in server
- docs(console): Update docstrings
- feat(doc): Handle nested reexports
Examples of bad title:
- fix #7123
- update docs
- fix bugs
2. Ensure there is a related issue and it is referenced in the PR text.
3. Ensure there are tests that cover the changes.
4. Ensure `cargo test` passes.
5. Ensure `./tools/format.js` passes without changing files.
6. Ensure `./tools/lint.js` passes.
7. Open as a draft PR if your work is still in progress. The CI won't
run
all steps, but you can add '[ci]' to a commit message to force it to.
8. If you would like to run the benchmarks on the CI, add the 'ci-bench'
label.
-->
## WHY
ref: https://github.com/denoland/deno/issues/19165
Node's fs/promises includes a FileHandle class, but deno does not. The
open function in Node's fs/promises returns a FileHandle, which provides
an IO interface to the file. However, deno's open function returns a
resource id.
### deno
```js
> const fs = await import("node:fs/promises");
undefined
> const file3 = await fs.open("./README.md");
undefined
> file3
3
> file3.read
undefined
Node:
```
### Node
```js
> const fs = await import("fs/promises");
undefined
> const file3 = await fs.open("./tests/e2e_unit/testdata/file.txt");
undefined
> file3
FileHandle {
_events: [Object: null prototype] {},
_eventsCount: 0,
_maxListeners: undefined,
close: [Function: close],
[Symbol(kCapture)]: false,
[Symbol(kHandle)]: FileHandle {},
[Symbol(kFd)]: 24,
[Symbol(kRefs)]: 1,
[Symbol(kClosePromise)]: null
}
> file3.read
[Function: read]
```
To be compatible with Node, deno's open function should also return a
FileHandle.
## WHAT
I have implemented the first step in adding a FileHandle.
- Changed the return value of the open function to a FileHandle object
- Implemented the readFile method in FileHandle
- Add test code
## What to do next
This PR is the first step in adding a FileHandle, and there are things
that should be done next.
- Add functionality equivalent to Node's FileHandle to FileHandle
(currently there is only readFile)
---------
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
For the first implementation of node:http2, we'll use the internal
version of `Deno.serve` which allows us to listen on a raw TCP
connection rather than a listener.
This is mostly a refactoring, and hooking up of `op_http_serve_on` that
was never previously exposed (but designed for this purpose).
This PR adds the missing `process.reallyExit()` method to node's
`process` object.
Was [pinged on
twitter](https://twitter.com/biwanczuk/status/1663326659787862017)
regarding running the `fastify` test suite in node. They use `node-tap`
which has been around arguably the longest of the test frameworks and
relies on a couple of old APIs. They have `signal-exit` as a dependency
which in turn [makes use of
`process.reallyExit()`](8fa7fc9a9c/src/index.ts (L19)).
That function cannot be found anywhere in their documentation, but
exists at runtime. See
6a6b3c5402/lib/internal/bootstrap/node.js (L172)
This doesn't yet make `node-tap` work, but gets us one step closer.
Part of #17840
I haven't made any changes or additions to the tests themselves, just
moved the tests over and updated to match.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
Rather than disallowing `ext:` resolution, clear the module map after
initializing extensions so extension modules are anonymized. This
operation is explicitly called in `deno_runtime`. Re-inject `node:`
specifiers into the module map after doing this.
Fixes #17717.
If a symlink within the `node_modules` directory lies outside that
directory, it will now warn and inline the file. For directories, it
will just warn for now.
Probably fixes #19251 (I'm still unable to reproduce).
`isFile`, `isDirectory`, `isSymlink` are defined in `Deno.FileInfo`, but
`isBlockDevice`, `isCharacterDevice`, `isFIFO`, `isSocket` are not
defined.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
This is what pnpm does and we were missing it. It makes modules work
which have a dependency on something, but don't say they have that
dependency, but that dep is still in the tree somewhere.
This commit fixes emitting "unhandledrejection" event when there are
"node:" or "npm:" imports.
Before this commit the Node "unhandledRejection" event was emitted
using a regular listener for Web "unhandledrejection" event. This
listener was installed before any user listener had a chance to be
installed which effectively prevent emitting "unhandledrejection"
events to user code.
Closes https://github.com/denoland/deno/issues/16928
When someone explicitly opts into using the node_modules dir via
`--node-modules-dir` or setting `"nodeModulesDir": true` in the
deno.json file, we should eagerly ensure a top level package.json
install is done on startup. This was initially always done when we added
package.json support and a package.json was auto-discovered, but scaled
it back to be lazily done when a bare specifier matched an entry in the
package.json because of how disruptive it was for people using Deno
scripts in Node projects. That said, it does not make sense for someone
to opt-into having deno control and use their node_modules directory and
not want a package.json install to occur. If such a rare scenario
exists, the `DENO_NO_PACKAGE_JSON=1` environment variable can be set.
Ideally, we would only ever use a node_modules directory with this
explicit opt-in so everything is very clear, but we still have this
automatic scenario when there's a package.json in order to make more
node projects work out of the box.
Noticed that we're checking more module paths than necessary. In
particular the module path array contains a couple of entries with a
duplicated `node_modules/node_modules` suffix.
```js
[
// ... more entries before here, where some also contain duplicate suffixes
"/Users/marvinhagemeister/dev/preact-render-to-string/node_modules/.deno/node_modules",
"/Users/marvinhagemeister/dev/preact-render-to-string/node_modules/node_modules", // <-- duplicate suffix
"/Users/marvinhagemeister/dev/preact-render-to-string/node_modules",
"/Users/marvinhagemeister/dev/node_modules",
"/Users/marvinhagemeister/node_modules",
"/Users/node_modules",
"/node_modules",
"/node_modules" // <-- duplicate entry
]
```
This was caused by a misunderstanding in how Rust's
[`Path::ends_with()`](https://doc.rust-lang.org/std/path/struct.Path.html#method.ends_with)
works. It's designed to match on whole path segments and the suffix
`/node_modules` is not that, except for the root entry. This meant that
our check for if the path already ended with `node_module` always
returned `false`. Removing the leading slash fixes that.
While we're at it, we can remove the last condition where we explicitly
added the root `/node_modules` entry since the while loop prior to that
takes care of it already.
Follow up to https://github.com/denoland/deno/pull/19084.
This commit adds support for globs in the configuration file as well
as CLI arguments for files.
With this change users can now use glob syntax for "include" and
"exclude" fields, like so:
```json
{
"lint": {
"include": [
"directory/test*.ts",
"other_dir/"
],
"exclude": [
"other_dir/foo*.ts",
"nested/nested2/*"
]
},
"test": {
"include": [
"data/test*.ts",
"nested/",
"tests/test[1-9].ts"
],
"exclude": [
"nested/foo?.ts",
"nested/nested2/*"
]
}
}
```
Or in CLI args like so:
```
// notice quotes here; these values will be passed to Deno verbatim
// and deno will perform glob expansion
$ deno fmt --ignore="data/*.ts"
$ deno lint "data/**/*.ts"
```
Closes https://github.com/denoland/deno/issues/17971
Closes https://github.com/denoland/deno/issues/6365
This adds support for the lockfile and node_modules directory to the
lsp.
In the case of the node_modules directory, it is only enabled when
explicitly opted into via `"nodeModulesDir": true` in the configuration
file. This is to reduce the language server automatically modifying the
node_modules directory when the user doesn't want it to.
Closes #16510
Closes #16373
This runs our `js_unit_tests` and `node_unit_tests` in parallel, one
rust test per JS unit test file. Some of our JS tests don't like running
in parallel due to port requirements, so this also makes those use a
specific port-per-file. This does not attempt to make the node-compat
tests work.
We don't need to use the `deno` command here to test kill permissions
and it's awkward to get right without passing `-A`. `cat` works, but for
platforms other than windows. This test should have plenty of coverage
on other platforms.
This commit changes the implementation of `ext/web` timers, by using
"op_void_async_deferred" for timeouts of 0ms.
0ms timeout is meant to be run at the end of the event loop tick and
currently Tokio timers that we use to back timeouts have at least 1ms
resolution. That means that 0ms timeout actually take >1ms. This
commit changes that and runs 0ms timeout at the end of the event
loop tick.
One consequence is that "unrefing" a 0ms timer will actually keep
the event loop alive (which I believe actually makes sense, the test
we had only worked because the timeout took more than 1ms).
Ref https://github.com/denoland/deno/issues/19034
We never properly added support for this. This fixes vendoring when it
has npm or node specifiers. Vendoring occurs by adding a
`"nodeModulesDir": true` property to deno.json then it uses a local
node_modules directory. This can be opted out by setting
`"nodeModulesDir": false` or running with `--node-modules-dir=false`.
Closes #18090
Closes #17210
Closes #17619
Closes #16778
This commit changes the return type of an unstable `Deno.serve()` API
to instead return a `Deno.Server` object that has a `finished` field.
This change is done in preparation to be able to ref/unref the HTTP
server.
This is a bit bare bones but gets `npm-run-all` working. For full stdio
compatibility with node more work is needed which is probably better
done in follow up PRs.
Fixes #19159
Note: If the package information has already been cached, then this
requires running with `--reload` or for the registry information to be
fetched some other way (ex. the cache busting).
Closes #15544
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
This commit reimplements most of "node:http" client APIs using
"ext/fetch".
There is some duplicated code and two removed Node compat tests that
will be fixed in follow up PRs.
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
Fixes for various `Attemped to access invalid request` bugs (#19058,
#15427, #17213).
We did not wait for both a drop event and a completion event before
removing items from the slab table. This ensures that we do so.
In addition, the slab methods are refactored out into `slab.rs` for
maintainability.
Currently the `multipart/form-data` parser in
`Request.prototype.formData` and `Response.prototype.formData` decodes
non-ASCII filenames incorrectly, as if they were encoded in Latin-1
rather than UTF-8. This happens because the header section of each
`multipart/form-data` entry is decoded as Latin-1 in order to be parsed
with `Headers`, which only allows `ByteString`s, but the names and
filenames are never decoded correctly. This PR fixes this as a
post-processing step.
Note that the `multipart/form-data` parsing for this APIs in the Fetch
spec is very much underspecified, and it does not specify that names and
filenames must be decoded as UTF-8. However, it does require that the
bodies of non-`File` entries are decoded as UTF-8, and in browsers,
names and filenames always use the same encoding as the body.
Closes #19142.
Partially supersedes #19016.
This migrates `spawn` and `spawn_blocking` to `deno_core`, and removes
the requirement for `spawn` tasks to be `Send` given our single-threaded
executor.
While we don't need to technically do anything w/`spawn_blocking`, this
allows us to have a single `JoinHandle` type that works for both cases,
and allows us to more easily experiment with alternative
`spawn_blocking` implementations that do not require tokio (ie: rayon).
Async ops (+~35%):
Before:
```
time 1310 ms rate 763358
time 1267 ms rate 789265
time 1259 ms rate 794281
time 1266 ms rate 789889
```
After:
```
time 956 ms rate 1046025
time 954 ms rate 1048218
time 924 ms rate 1082251
time 920 ms rate 1086956
```
HTTP serve (+~4.4%):
Before:
```
Running 10s test @ http://localhost:4500
2 threads and 10 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 68.78us 19.77us 1.43ms 86.84%
Req/Sec 68.78k 5.00k 73.84k 91.58%
1381833 requests in 10.10s, 167.36MB read
Requests/sec: 136823.29
Transfer/sec: 16.57MB
```
After:
```
Running 10s test @ http://localhost:4500
2 threads and 10 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 63.12us 17.43us 1.11ms 85.13%
Req/Sec 71.82k 3.71k 77.02k 79.21%
1443195 requests in 10.10s, 174.79MB read
Requests/sec: 142921.99
Transfer/sec: 17.31MB
```
Suggested-By: alice@ryhl.io
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
Adds a `deno.preloadLimit` option (ex. `"deno.preloadLimit": 2000`)
which specifies how many file entries to traverse on the file system
when the lsp loads or its configuration changes.
Closes #18955
This PR ensures that node's `worker_threads` module exports
`MessageChannel`, `MessagePort` and the `BroadcastChannel` API. Fixing
these won't make `esbuild` work, but brings us one step closer 🎉
Fixes #19028 .
This is the initial support for npm and node specifiers in `deno
compile`. The npm packages are included in the binary and read from it via
a virtual file system. This also supports the `--node-modules-dir` flag,
dependencies specified in a package.json, and npm binary commands (ex.
`deno compile --unstable npm:cowsay`)
Closes #16632
`Content-Encoding: gzip` support for `Deno.serve`. This doesn't support
Brotli (`br`) yet, however it should not be difficult to add. Heuristics
for compression are modelled after those in `Deno.serveHttp`.
Tests are provided to ensure that the gzip compression is correct. We
chunk a number of different streams (zeros, hard-to-compress data,
already-gzipped data) in a number of different ways (regular, random,
large/small, small/large).
Fixes #16699 and #18960 by ensuring that we release our HTTP
`spawn_local` tasks when the HTTP resource is dropped.
Because our cancel handle was being projected from the resource via
`RcMap`, the resource was never `Drop`ped. By splitting the handle out
into its own `Rc`, we can avoid keeping the resource alive and let it
drop to cancel everything.
Fixes #18979.
This changes the predicate for allowing `ext:` specifier resolution from
`snapshot_loaded_and_not_snapshotting` to `ext_resolution_allowed` which
is only set to true during the extension module loading phase. Module
loaders as used in core
are now declared as `ExtModuleLoader` rather than `dyn ModuleLoader`.
`new Deno.KvU64(1n) + 2n == 3n` is now true.
`new Deno.KvU64(1n)` is now inspected as `[Deno.KvU64: 1n]`
(`Object(1n)` is inspected as `[BigInt: 1n]`).
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
This commit changes how paths for npm packages are handled,
by canonicalizing them when resolving. This is done so that instead
of returning
"node_modules/<package_name>@<version>/node_modules/<dep>/index.js"
(which is a symlink) we "node_modules/<dep>@<dep_version>/index.js.
Fixes https://github.com/denoland/deno/issues/18924
Fixes https://github.com/bluwy/create-vite-extra/issues/31
---------
Co-authored-by: David Sherret <dsherret@gmail.com>
This removes `ProcState` and replaces it with a new `CliFactory` which
initializes our "service structs" on demand. This isn't a performance
improvement at the moment for `deno run`, but might unlock performance
improvements in the future.
- Do not use `ReflectHas` in `isNode`.
- Avoid copying handler array when handlers.length == 1
- Avoid searching for path target when path.length == 1
```
Linux divy-2 5.19.0-1022-gcp #24~22.04.1-Ubuntu SMP Sun Apr 23 09:51:08 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
32GiB System memory
Intel(R) Xeon(R) CPU @ 3.10GHz
# main + https://github.com/denoland/deno/pull/18904
Msg/sec: 89326.750000
Msg/sec: 90320.000000
Msg/sec: 89576.250000
# this patch
Msg/sec: 97250.000000
Msg/sec: 97125.500000
Msg/sec: 97964.500000
```
Fixes regression from #18878 where `Promise.reject()`,
`Promise.reject(undefined)` and `reportError(undefined)` panic in the
REPL.
Fixes `throw undefined` printing `Uncaught Unknown exception` instead of
`Uncaught undefined`.
Fixes #8858.
Fixes #8869.
```
$ target/debug/deno
Deno 1.32.5
exit using ctrl+d, ctrl+c, or close()
REPL is running with all permissions allowed.
To specify permissions, run `deno repl` with allow flags.
> Promise.reject(new Error("bar"));
Promise { <rejected> Error: bar
at <anonymous>:2:16 }
Uncaught (in promise) Error: bar
at <anonymous>:2:16
> reportError(new Error("baz"));
undefined
Uncaught Error: baz
at <anonymous>:2:13
>
Towards https://github.com/denoland/deno/issues/18455
`safe`, `add` and `rem` options are not implemented because there is no
rust crate that provides this functionality (except rust-openssl maybe)
and its just not clear if this API is used widely.
This implements HTTP/2 prior-knowledge connections, allowing clients to
request HTTP/2 over plaintext or TLS-without-ALPN connections. If a
client requests a specific protocol via ALPN (`h2` or `http/1.1`),
however, the protocol is forced and must be used.
This is a rewrite of the `Deno.serve` API to live on top of hyper
1.0-rc3. The code should be more maintainable long-term, and avoids some
of the slower mpsc patterns that made the older code less efficient than
it could have been.
Missing features:
- `upgradeHttp` and `upgradeHttpRaw` (`upgradeWebSocket` is available,
however).
- Automatic compression is unavailable on responses.
1. Adds cli/standalone folder
2. Writes the bytes directly to the output file. When adding npm
packages this might get quite large, so let's not keep the final output
in memory just in case.
Towards #18455
This commit implements the keypair generation for asymmetric keys for
the `generateKeyPair` API.
See how key material is managed in this implementation:
https://www.notion.so/denolandinc/node-crypto-design-99fc33f568d24e47a5e4b36002c5325d?pvs=4
Private and public key encoding depend on `KeyObject#export` which is
not implemented. I've also skipped ED448 and X448 since we need a crate
for that in WebCrypto too.
Fixes https://github.com/denoland/deno/issues/18700
Timeline of the events that lead to the bug.
1. WebSocket handshake complete
2. Server on `read_frame` holding an AsyncRefCell borrow of the
WebSocket stream.
3. Client sends a TXT frame after a some time
4. Server recieves the frame and goes back to `read_frame`.
5. After some time, Server starts a `write_frame` but `read_frame` is
still holding a borrow!
^--- Locked. read_frame needs to complete so we can resume the write.
This commit changes all writes to directly borrow the
`fastwebsocket::WebSocket` resource under the assumption that it won't
affect ongoing reads.
This commit refactors "deno_core" to do fewer boundary crossings
from Rust to V8. In other words we are now calling V8 from Rust fewer
times.
This is done by merging 3 distinct callbacks into a single one. Instead
of having "op resolve" callback, "next tick" callback and "macrotask
queue" callback, we now have only "Deno.core.eventLoopTick" callback,
which is responsible for doing the same actions previous 3 callbacks.
On each of the event loop we were doing at least 2 boundary crosses
(timers macrotask queue callback and unhandled promise rejection
callback) and up to 4 crosses if there were op response and next tick
callbacks coming from Node.js compatibility layer. Now this is all done
in a single callback.
Closes https://github.com/denoland/deno/issues/18620
Stores the test/bench functions in rust op state during registration.
The functions are wrapped in JS first so that they return a directly
convertible `TestResult`/`BenchResult`. Test steps are still mostly
handled in JS since they are pretty much invoked by the user. Allows
removing a bunch of infrastructure for communicating between JS and
rust. Allows using rust utilities for things like shuffling tests
(`Vec::shuffle`). We can progressively move op and resource sanitization
to rust as well.
Fixes #17122.
Fixes #17312.
Fixes #6259.
Adds the location for v8 syntax errors to the message (`message += " at
{location}"`) when rethrowing them for dynamic imports.
Discussing with @bartlomieju on discord I proposed just preserving v8's
error and not reconstructing it, allowing the standard stack trace to
just point to the syntax error instead of the dynamic import. But on
further thought this way has parity with SWC's syntax errors + has the
advantage of showing both the syntax error and dynamic import location.
```ts
// temp.js
await import("./temp2.js");
// temp2.js
function foo() {
await Promise.resolve();
}
// Before:
// error: Uncaught (in promise) SyntaxError: Unexpected reserved word
// await import("./temp2.js");
// ^
// at async file:///.../temp.js:1:1
// After:
// error: Uncaught (in promise) SyntaxError: Unexpected reserved word at file:///.../temp2.js:2:3
// await import("./temp2.js");
// ^
// at async file:///.../temp.js:1:1
```
This reloads an npm package's dependency's information when a
version/version req/tag is not found.
This PR applies only to dependencies of npm packages. It does NOT yet
cause npm specifiers to have their dependency information cache busted.
That requires a different solution, but this should help cache bust in
more scenarios.
Part of #16901, but doesn't close it yet
Turns out `autoprefixer` is a better reproduction case then
`microbundle`.
Fixes #18535
Fixes #18600
---------
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
This commit updates the `Deno.Kv` API to return the new commited
versionstamp for the mutated data from `db.set` and `ao.commit`. This is
returned in the form of a `Deno.KvCommitResult` object that has a
`versionstamp` property.
This reverts commit a3529d0232.
This change made debugging Node tests very hard - `AssertionError` is
now printed as `[Circular *1]` giving no visibility what failed.
We need to align two implementations together and remove this one then.
1. Fixes a cosmetic issue in the repl where it would display lsp warning
messages.
2. Lazily loads dependencies from the package.json on use.
3. Supports using bare specifiers from package.json in the REPL.
Closes #17929
Closes #18494
We currently run the all test cases in `parallel` category at the same
time, which invokes hundreds process at the same time, and that seems
causing some flakiness in CI. (maybe related to #18487)
This PR limits the concurrency to the number of cpu cores. This is more
aligned to how Node.js run their `parallel` test in their repository.
42c4a35952/Makefile (L356)
1. Rewrites the tests to be more back and forth rather than getting the
output all at once (which I believe was causing the hangs on linux and
maybe mac)
2. Runs the pty tests on the linux ci.
3. Fixes a bunch of tests that were just wrong.
4. Adds timeouts on the pty tests.
Towards #18455
This commit implements `checkPrimeSync` and `checkPrime` in node:crypto
using the Miller-Rabin primality test (fun fact: it actually is a test
for composite numbers)
It first compares the candidate against many known small primes and if
not, proceeds to run the Miller-Rabin primality test.
http://nickle.org/examples/miller-rabin.5c used as reference
implementation.
This gets SQLite off the flamegraph and reduces initialization time by
somewhere between 0.2ms and 0.5ms. In addition, I took the opportunity
to move all the cache management code to a single place and reduce
duplication. While the PR has a net gain of lines, much of that is just
being a bit more deliberate with how we're recovering from errors.
The existing caches had various policies for dealing with cache
corruption, so I've unified them and tried to isolate the decisions we
make for recovery in a single place (see `open_connection` in
`CacheDB`). The policy I chose was:
1. Retry twice to open on-disk caches
2. If that fails, try to delete the file and recreate it on-disk
3. If we fail to delete the file or re-create a new cache, use a
fallback strategy that can be chosen per-cache: InMemory (temporary
cache for the process run), BlackHole (ignore writes, return empty
reads), or Error (fail on every operation).
The caches all use the same general code now, and share the cache
failure recovery policy.
In addition, it cleans up a TODO in the `NodeAnalysisCache`.
This internal node hook is used by libraries such as `ts-node` when used
as a require hook `node -r ts-node/register`. That combination is often
used with test frameworks like `mocha` or `jasmine`.
We had a reference to `Module._preloadModules` in our code, but the
implementation was missing. While fixing this I also noticed that the
`fakeParent` module that we create internally always threw because of
the `pathDirname` check on the module id in the constructor of `Mdoule`.
So this code path was probably broken for a while.
```txt
✖ ERROR: Error: Empty filepath.
at pathDirname (ext:deno_node/01_require.js:245:11)
at new Module (ext:deno_node/01_require.js:446:15)
at Function.Module._resolveFilename (ext:deno_node/01_require.js:754:28)
at Function.resolve (ext:deno_node/01_require.js:1015:19)
```
