v13.5.0 proposal by MylesBorins · Pull Request #31010 · nodejs/node

@joaocgreis @MylesBorins

Explicitly pass the WiX SDK directory when building the MSI. WiX
doesn't (yet?) have a directory for VS2019, so use the one for VS2017
which should be compatible.

PR-URL: #30895
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@BridgeAR @MylesBorins

This makes sure that the regular expression matches all built-in
objects properly. So far a couple where missed.

PR-URL: #30768
Fixes: #30183
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>

@BridgeAR @MylesBorins

This is only active if the `showHidden` option is truthy.

The implementation is a trade-off between accuracy and performance.
This will miss properties such as properties added to built-in data
types.

The goal is mainly to visualize prototype getters and setters such as:

class Foo {
  ownProperty = true
  get bar() {
    return 'Hello world!'
  }
}

const a = new Foo()

The `bar` property is a non-enumerable property on the prototype while
`ownProperty` will be set directly on the created instance.

The output is similar to the one of Chromium when inspecting objects
closer. The output from Firefox is difficult to compare, since it's
always a structured interactive output and was therefore not taken
into account.

PR-URL: #30768
Fixes: #30183
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>

@Sebastien-Ahkrin @MylesBorins

PR-URL: #30908
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@dev-script @MylesBorins

doc change for stream.md that 'finish' event should be before
writer.end

fixes: #30759

PR-URL: #30889
Fixes: #30759
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@Sebastien-Ahkrin @MylesBorins

PR-URL: #30905
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@targos @MylesBorins

This fixes two issues in the REPL when it is started with a new context
(useGlobal option set to `false`):
- The `primordials` object does not contain all builtins, so the
  filtering based on property names from `primordials` was wrong.
- The autocompleter did not take builtin names into account because
  they are not properties of the context object.

A list of all global builtin names is created lazily when needed. It is
used for filtering for the copy and for adding those names to the
autocompleter list.

Fixes: #30792

PR-URL: #30883
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: David Carlier <devnexen@gmail.com>

@legendecas @MylesBorins

PR-URL: #30877
Reviewed-By: David Carlier <devnexen@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@joyeecheung @MylesBorins

Refactor the internal NativeModule class to a JS class and add
more documentation about its properties.

PR-URL: #30856
Reviewed-By: Denys Otrishko <shishugi@gmail.com>
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@sgallagher @MylesBorins

When compiled with `--with-intl=small` and
`--with-icu-default-data-dir=PATH`, Node.js will use PATH as a
fallback location for the ICU data.

We will first perform an access check using fopen(PATH, 'r') to
ensure that the file is readable. If it is, we'll set the
icu_data_directory and proceed. There's a slight overhead for the
fopen() check, but it should be barely measurable.

This will be useful for Linux distribution packagers who want to
be able to ship a minimal node binary in a container image but
also be able to add on the full i18n support where needed. With
this patch, it becomes possible to ship the interpreter as
/usr/bin/node in one package for the distribution and to ship the
data files in another package (without a strict dependency
between the two). This means that users of the distribution will
not need to explicitly direct Node.js to locate the ICU data. It
also means that in environments where full internationalization is
not required, they do not need to carry the extra content (with
the associated storage costs).

Refs: #3460

Signed-off-by: Stephen Gallagher <sgallagh@redhat.com>

PR-URL: #30825
Reviewed-By: Steven R Loomis <srloomis@us.ibm.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@ronag @MylesBorins

PR-URL: #30852
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@PW486 @MylesBorins

Change '==' to '===' in v8_prof_polyfill.js, punycode.js.

PR-URL: #30898
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: David Carlier <devnexen@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@timcosta @MylesBorins

Fixes: #24980
Refs: eb43bc04b1

PR-URL: #30071
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Trivikram Kamat <trivikr.dev@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@apapirovski @MylesBorins

process.nextTick accepts additional parameters which
are passed through to the callback. Use that instead
of binding the function to a context.

PR-URL: #28131
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

@apapirovski @MylesBorins

Pass through parameters using setImmediate rather
than using Function.prototype.bind to bind the
provided context.

PR-URL: #28131
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

@apapirovski @MylesBorins

Don't use Function.prototype.bind where it isn't
necessary. Rely on event emitter context instead
and on arrow function as class property.

PR-URL: #28131
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

@apapirovski @MylesBorins

PR-URL: #28131
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

@apapirovski @MylesBorins

PR-URL: #28131
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>

@codebytere @MylesBorins

This simplifies requires for those using DebugOptions,
since debug_options was defined in src/node_options-inl.h  and thus
embedders would need to require an extra file to do what could
trivially be consolidated into one.

PR-URL: #30494
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: David Carlier <devnexen@gmail.com>

@mcollina @MylesBorins

This changes makes all stream in a pipeline emit 'error' in
case of an abnormal termination of the pipeline. If the last stream
is currently being async iterated, this change will make the iteration
reject accordingly.

See: #30861
Fixes: #28194

PR-URL: #30869
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@codebytere @MylesBorins

OpenSSL uses a macro without typechecking; since C++ does not
implicitly cast void* this is needed to conform Node.js to the
OpenSSL documentation.

PR-URL: #30917
Reviewed-By: Tobias Nießen <tniessen@tnie.de>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@Trott @MylesBorins

There is a test that doesn't load the common module initially because it
needs to monkey-patch the 'os' module. I think it would be a good idea
to minimize the side-effects of loading common anyway, so let's defer
loading 'os' unless/until it's actually needed.

PR-URL: #30914
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>

@Trott @MylesBorins

Most tests in `test/parallel` work when invoked with `node` rather than
`tools/test.py` but not test-os-checked-function because it doesn't load
the `common` module initially, which means it won't get re-spawned with
the necessary flags (in the Flags: comment, in this case
--expose_internals). Now that common delays loading 'os' until it needs
to load it, this test can load the common module and it will work from
the command line without the test harness. Additionally, we now can
remove a comment disabling a lint rule.

PR-URL: #30914
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>

@Trott @MylesBorins

PR-URL: #30913
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>

@MylesBorins

PR-URL: #30910
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Gireesh Punathil <gpunathi@in.ibm.com>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Trivikram Kamat <trivikr.dev@gmail.com>

@tniessen @MylesBorins

The return value is not a boolean and even if interpreted as one,
it does not indicate whether an exception is pending.

For napi_is_exception_pending, the description of the result parameter
already explains how to check whether an exception is pending.

PR-URL: #30893
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Gabriel Schulhof <gabriel.schulhof@intel.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>

@mcollina @MylesBorins

PR-URL: #30912
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@addaleax @MylesBorins

I previously thought the order of these calls was no longer
relevant. I was wrong.

This commit undoes the changes from 312c02d, adds a comment
explaining why I was wrong, and flips the order of the calls
elsewhere for consistency, the latter having been the goal
of 312c02d.

Fixes: #30846
Refs: #30181

PR-URL: #30909
Reviewed-By: Shelley Vohr <codebytere@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: David Carlier <devnexen@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>

@Trott @MylesBorins

PR-URL: #30919
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>

@cjihrig @MylesBorins

test-wasi-start-validation.js should require the
--experimental-wasi-unstable-preview0 flag in order to run.
However, due to a recent regression, that hasn't been enforced.
#30963 fixes the regression
and will cause this test to start (correctly) failing. This
commit adds the missing flag.

PR-URL: #30971
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>