If a readable stream was set up with `highWaterMark 0`, the while-loop
in `maybeReadMore_` function would never execute.
The while loop now has an extra or-condition for the case where the
stream is flowing and there are no items. The or-condition is adapted
from the emit-condition of the `addChunk` function.
The `addChunk` also contains a check for `state.sync`. However that part
of the check was omitted here because the `maybeReadMore_` is executed
using `process.nextTick`. `state.sync` is set and then unset within the
`read()` function so it should never be in effect in `maybeReadMore_`.
Fixes: https://github.com/nodejs/node/issues/24915
PR-URL: https://github.com/nodejs/node/pull/24918
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
When the worker thread exits, drain the messages also from the internal
message port so that the call to 'kDispose' will occur only after all
the messages from the worker were processed in the parent, so stdio
messages from the worker will be successfully pushed to their target
streams in the parent.
PR-URL: https://github.com/nodejs/node/pull/24932
Fixes: https://github.com/nodejs/node/issues/24636
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Currently it is not possible to inspect getters. To prevent any side
effects this should not become a default but under lots of
circumstances it would still be useful to inspect getters. This way
it is possible to actively opt into inspecting those.
PR-URL: https://github.com/nodejs/node/pull/24852
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
HTTP/2 streams call `.end()` on themselves from their
`.destroy()` method, which might be queued (e.g. due to network
congestion) and not processed before the stream itself is destroyed.
In that case, the `_writableState.ended` property could be set before
the stream emits its `'close'` event, and never actually emits the
`'finished'` event, confusing the end-of-stream implementation so
that it wouldn’t call its callback.
This can be fixed by watching for the end events themselves using the
existing `'finish'` and `'end'` listeners rather than relying on the
`.ended` properties of the `_...State` objects.
These properties still need to be checked to know whether stream
closure was premature – My understanding is that ideally, streams
should not emit `'close'` before `'end'` and/or `'finished'`, so this
might be another bug, but changing this would require modifying tests
and almost certainly be a breaking change.
Fixes: https://github.com/nodejs/node/issues/24456
PR-URL: https://github.com/nodejs/node/pull/24926
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Franziska Hinkelmann <franziska.hinkelmann@gmail.com>
Instead of treating config.gypi as a JavaScript file, specialize
the processing in js2c and make the serialized result a real JSON
string (with 'true' and 'false' converted to boolean values) so
we don't have to use a custom deserializer during bootstrap.
In addition, store the JSON string separately in NativeModuleLoader,
and keep it separate from the map of the builtin source code, so
we don't have to put it onto `NativeModule._source` and delete it
later, though we still preserve it in `process.binding('natives')`,
which we don't use anymore.
This patch also makes the map of builtin source code and the
config.gypi string available through side-effect-free getters
in C++.
PR-URL: https://github.com/nodejs/node/pull/24816
Reviewed-By: Gus Caplan <me@gus.host>
In case of an error where we only care about a cleaned up stack
trace it is cheaper to reset the stack trace limit for the error
that is created. That way the stack frames do not have to be
computed twice.
PR-URL: https://github.com/nodejs/node/pull/24747
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
The path module's assertPath() does exactly what the
validateString() validator does, so this commit updates
path to use validateString() instead. A couple drive by
updates to validateString() outside of assertPath() are
also included.
PR-URL: https://github.com/nodejs/node/pull/24840
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
The only known condition where we could not provide appropriate
stdio streams so far were non-console Windows applications.
Since this issue has come up a few times in our issue tracker now,
switch to providing dummy streams for these cases instead.
If there are other valid cases in which `uv_guess_handle` fails,
and where there is a more sensible way to provide stdio,
we’ll probably still find out because the streams don’t work
properly either way.
Refs: https://github.com/nodejs/help/issues/1251
PR-URL: https://github.com/nodejs/node/pull/20640
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
This patch moves all the bootstrapper compilation to use
NativeModuleLoader::CompileAndCall(). With this we no longer
need to mess with the error decoration and handling any more -
there is no point in handling the JS error occurred during bootstrapping
by ourselves, we should just crash or let the VM handle it.
PR-URL: https://github.com/nodejs/node/pull/24775
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Comparing sparse arrays did not work properly. That is fixed and
tests were added to verify that everything works as expected.
This had an impact on `util.isDeepStrictEqual()` and
`assert.deepStrictEqual()` and their counterpart
`assert.notDeepStrictEqual()`.
PR-URL: https://github.com/nodejs/node/pull/24749
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
The current internal isError function checked the toString value
instead of using the more precise `util.types.isNativeError()` check.
The `instanceof` check is not removed due to possible errors that
are not native but still an instance of Error.
PR-URL: https://github.com/nodejs/node/pull/24746
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
The `console` functions rely on the `util.format()` behavior. It
did not follow the whatwg spec when it comes to symbols in combination
with the %d, %i and %f format specifiers. Using a symbol argument in
combination with one of these specifiers resulted in an error instead
of returning `'NaN'`. This is now fixed by this patch.
PR-URL: https://github.com/nodejs/node/pull/23708
Refs: https://console.spec.whatwg.org/#formatter
Reviewed-By: Roman Reiss <me@silverwind.io>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Move the wrapping of the inspector console in a separate file
for clarity. In addition, save the original console from the
VM explicitly via an exported property
`require('internal/console/inspector').consoleFromVM`
that `require('inspector').console` can alias to it later,
instead of hanging the original console onto `per_thread.js`
during bootstrap and counting on that `per_thread.js`
only gets evaluated once and gets cached.
PR-URL: https://github.com/nodejs/node/pull/24709
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
Since we do not actually use the Console constructor to
instantiate the global console, move the two piece of
code into two different JS files for clarity, and make
console.js a mere re-export of the global console.
The hope is to make the global console, a namespace, more
web-compatible while keeping the Console constructor
available for backwards compatibility.
PR-URL: https://github.com/nodejs/node/pull/24709
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
