Main change is that:
- "hyper" has been renamed to "hyper_v014" to signal that it's legacy
- "hyper1" has been renamed to "hyper" and should be the default
`opAsync` requires a lookup by name on each async call. This is a
mechanical translation of all opAsync calls to ensureFastOps.
The `opAsync` API on Deno.core will be removed at a later time.
When we migrate to op-import-per-extension, we will want to ensure that
ops have one and only one place where they are imported. This tackles
the ops that are imported via `ensureFastOps`, but does not yet tackle
direct `ops` imports.
Landing ahead of https://github.com/denoland/deno_core/pull/393
Node HTTP/2 was using the default h2 `Bytes` datatype when we can be
making using of `BufView` like we do in `Deno.serve`.
`fetch` and `Deno.serverHttp` can't make use of `BufView` because they
are using `reqwest` which is stuck on hyper 0.x at this time.
This PR implements the child_process IPC pipe between parent and child.
The implementation uses Windows named pipes created by parent and passes
the inheritable file handle to the child.
I've also replace parts of the initial implementation which passed the
raw parent fd to JS with resource ids instead. This way no file handle
is exposed to the JS land (both parent and child).
`IpcJsonStreamResource` can stream upto 800MB/s of JSON data on Win 11
AMD Ryzen 7 16GB (without `memchr` vectorization)
Bumped versions for 1.39.0
Please ensure:
- [x] Target branch is correct (`vX.XX` if a patch release, `main` if
minor)
- [x] Crate versions are bumped correctly
- [x] deno_std version is incremented in the code (see
`cli/deno_std.rs`)
- [x] Releases.md is updated correctly (think relevancy and remove
reverts)
To make edits to this PR:
```shell
git fetch upstream release_1_39.0 && git checkout -b release_1_39.0 upstream/release_1_39.0
```
cc @mmastrac
---------
Co-authored-by: mmastrac <mmastrac@users.noreply.github.com>
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
This PR implements the Node child_process IPC functionality in Deno on
Unix systems.
For `fd > 2` a duplex unix pipe is set up between the parent and child
processes. Currently implements data passing via the channel in the JSON
serialization format.
Part 2 of removing middleware.
This is somewhat awkward because `V8CrossThreadTaskSpawner` requires
tasks to be `Send`, but NAPI makes heavy use of `!Send` pointers. In
addition, Rust causes a closure to be `!Send` if you pull a `!Send`
value out of a struct.
---------
Signed-off-by: Matt Mastracci <matthew@mastracci.com>
Co-authored-by: Divy Srivastava <dj.srivastava23@gmail.com>
This problem occurred trying to load tensorflow.js
```
> import * as tf from 'npm:@tensorflow/tfjs';
Uncaught SyntaxError: Identifier 'mod' has already been declared at file:///Users/ry/Library/Caches/deno/npm/registry.npmjs.org/@tensorflow/tfjs/4.14.0/dist/tf.node.js:167:14
at async <anonymous>:1:33
```
This commit adds a no-op flushHeaders method to the ServerResponse
object. It is a nop because the ServerResponse implementation is based
on top of the Deno server API instead of the Node `OutgoingMessage`
base.
Fixes #21509
This significantly optimizes URLPattern in the case where the same
URL is matched against many patterns (like in a router).
Also minor speedups to other use-cases.
This PR is an attempt to fix
https://github.com/denoland/deno/issues/20293, in which node modules
connecting to databases fail due to TLS errors. I ran into this
attempting to use
[node-postgres](https://github.com/brianc/node-postgres) to connect to a
[Neon](https://neon.tech) database.
Investigating via `--inspect-brk` led me to notice that the hostname
eventually passed to `Deno.startTls` was null. The hostname is
determined by the following code:
f6b889b432/ext/node/polyfills/_tls_wrap.ts (L87-L89)
This logic doesn't appear to be correct. I couldn't find reference to
`servername` existing on the `secureContext` in either Node's or Deno's
docs. There's a lot of scope here, and it's my first time reading
through this code, so I could be missing something!
Node uses [the following
logic](2e458d9736/lib/_tls_wrap.js (L1679-L1682)
) to determine the hostname for certificate validation:
```
const hostname = options.servername ||
options.host ||
(options.socket && options.socket._host) ||
'localhost';
```
This PR updates the `TLSSocket` polyfill to use behave similarly (though
I omitted the default to `localhost` at the end; I'm not sure if
including it is necessary or correct). With this change, `node-postgres`
connects to my TLS endpoint successfully (aside: Neon requires SNI,
which also works as expected).
---
I tried to update the tests in
https://github.com/denoland/deno/blob/main/cli/tests/unit_node/tls_test.ts
to exercise this change, but the test fails for me on `main` on Linux. I
investigated briefly and noticed that the test fixture
`cli/tests/testdata/tls/localhost.crt` doesn't appear to include the
`subjectAltName` specified in `domains.txt`. I believe the certificate
isn't matching `localhost`, but that's where I ended investigating.
This commit refactors how we access "core", "internals" and
"primordials" objects coming from `deno_core`, in our internal JavaScript code.
Instead of capturing them from "globalThis.__bootstrap" namespace, we
import them from recently added "ext:core/mod.js" file.
Landing changes required for
https://github.com/denoland/deno_core/pull/359
We needed to update 99_main.js and a whole load of tests.
API changes:
- setPromiseRejectCallback becomes setUnhandledPromiseRejectionHandler.
The function is now called from eventLoopTick.
- The promiseRejectMacrotaskCallback no longer exists, as this is
automatically handled in eventLoopTick.
- ops.op_dispatch_exception now takes a second parameter: in_promise.
The preferred way to call this op is now reportUnhandledException or
reportUnhandledPromiseRejection.
This commit stabilizes "Deno.HttpServer.shutdown" API as well as
Unix socket support in "Deno.serve" API.
---------
Co-authored-by: Yoshiya Hinosawa <stibium121@gmail.com>
This commit adds a method of `Symbol.dispose` to the object returned
from `Deno.createHttpClient`, so we can make use of [explicit resource
management](https://github.com/tc39/proposal-explicit-resource-management)
by declaring it with `using`.
This commit adds support for a new `kv.watch()` method that allows
watching for changes to a key-value pair. This is useful for cases
where you want to be notified when a key-value pair changes, but
don't want to have to poll for changes.
---------
Co-authored-by: losfair <zhy20000919@hotmail.com>
Add support for signing with a RSA PEM private key: `pkcs8` and `pkcs1`.
Fixes https://github.com/denoland/deno/issues/18972
Ref #21124
Verified fix with `npm:sshpk`. Unverfied but fixes
`npm:google-auth-library`, `npm:web-push` & `oracle/oci-typescript-sdk`
---------
Signed-off-by: Divy Srivastava <dj.srivastava23@gmail.com>
Switch `ext/fetch` over to `resourceForReadableStream` to simplify and
unify implementation with `ext/serve`. This allows us to work in Rust
with resources only.
Two additional changes made to `resourceForReadableStream` were
required:
- Add an optional length to `resourceForReadableStream` which translates
to `size_hint`
- Fix a bug where writing to a closed stream that was full would panic
Rust 1.74 may have made this code temporarily valid in [#113126 Replace
old private-in-public diagnostic with type privacy
lints](https://github.com/rust-lang/rust/pull/113126), so we didn't
catch it at build time.
It fails in 1.73 and +nightly, however.
Follow-up to #20822. cc @lrowe
The `httpServerExplicitResourceManagement` tests were randomly failing
on CI because of a race.
The `drain` waker was missing wakeup events if the listeners shut down
after the last HTTP response finished. If we lost the race (rare), the
server Rc would be dropped and we wouldn't poll it again.
This replaces the drain waker system with a signalling Rc that always
resolves when the refcount is about to become 1.
Fix verified by running serve tests in a loop:
```
for i in {0..100}; do cargo run --features=__http_tracing -- test
-A --unstable '/Users/matt/Documents/github/deno/deno/cli/tests/unit/ser
ve_test.ts' --filter httpServerExplicitResourceManagement; done;
```
This PR changes the `Deno.cron` API:
* Marks the existing function as deprecated
* Introduces 2 new overloads, where the handler arg is always last:
```ts
Deno.cron(
name: string,
schedule: string,
handler: () => Promise<void> | void,
)
Deno.cron(
name: string,
schedule: string,
options?: { backoffSchedule?: number[]; signal?: AbortSignal },
handler: () => Promise<void> | void,
)
```
This PR also fixes a bug, when other crons continue execution after one
of the crons was closed using `signal`.
Fixes #21121 and #19498
Migrates fully to rustls_tokio_stream. We no longer need to maintain our
own TlsStream implementation to properly support duplex.
This should fix a number of errors with TLS and websockets, HTTP and
"other" places where it's failing.
Use HttpRecord as response body so requests can be tracked all the way
to response body completion.
This allows Request properties to be accessed while the response body is
streaming.
Graceful shutdown now awaits a future instead of async spinning waiting
for requests to finish.
On the minimal benchmark this refactor improves performance an
additional 2% over pooling alone for a net 3% increase over the previous
deno main branch.
Builds upon https://github.com/denoland/deno/pull/20809 and
https://github.com/denoland/deno/pull/20770.
---------
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
Reuse existing existing allocations for HttpRecord and response
HeaderMap where possible.
At request end used allocations are returned to the pool and the pool
and the pool sized to 1/8th the current number of inflight requests.
For http1 hyper will reuse the response HeaderMap for the following
request on the connection.
Builds upon https://github.com/denoland/deno/pull/20770
---------
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
Makes the JavaScript Request use a v8:External opaque pointer to
directly refer to the Rust HttpRecord.
The HttpRecord is now reference counted. To avoid leaks the strong count
is checked at request completion.
Performance seems unchanged on the minimal benchmark. 118614 req/s this
branch vs 118564 req/s on main, but variance between runs on my laptop
is pretty high.
---------
Co-authored-by: Matt Mastracci <matthew@mastracci.com>
This PR uses the new `cancel` method of `TransformStream` to properly
clean up the internal `TextDecoder` used in `TextDecoderStream` if the
stream is cancelled.
Fixes #13142
Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
