This patch improves the means by which the docs are viewed locally.
* Remove extraneous code in the `docserve` Makefile target
* Document the `docserve` task for all to know
* Bring all code snippets in this section up to speed
* Clarify the purpose of each documentation browsing method
Fixes: https://github.com/nodejs/node/issues/34977
PR-URL: https://github.com/nodejs/node/pull/34986
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
This adds a new make target to generate platform dependent
files for openssl on non-linux machines. The scripts we currently
have in place require linux. This adds a Dockerfile that installs
the necessary dependencies to be able to generate these files.
Previously, it was necessary to run `make -C deps/openssl/config`
on a linux machine. Now, as long as docker is installed and in
your `PATH`, it is possible to run `make gen-openssl`.
PR-URL: https://github.com/nodejs/node/pull/34642
Reviewed-By: James M Snell <jasnell@gmail.com>
Avoid building the node binary when building the source tarball. We
need a node binary to build the docs, but it doesn't have to be one
we build from scratch and can reuse any available node binary.
Skip building the xz compressed tarball in the build-tarball workflow
as we only use the gzip compressed tarball in the subsequent build
jobs.
PR-URL: https://github.com/nodejs/node/pull/34508
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: Zeyu Yang <himself65@outlook.com>
lint-js.js was implemented before ESLint had a caching feature. It is
now only used in CI. Let's remove it on the following grounds:
* It results in occasional (and puzzling) yellow CI runs for
node-test-linter because the tap file is corrupted somehow.
Interleaved maybe? I don't know, but a simple solution is removing it
and running ESLint directly.
* On my local laptop, it reduces the linting from about 75 seconds to
about 55 seconds. This kind of savings is not worth the added
complexity and the instability noted above.
PR-URL: https://github.com/nodejs/node/pull/30955
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>
Reviewed-By: Daijiro Wachi <daijiro.wachi@gmail.com>
Reviewed-By: Anatoli Papirovski <apapirovski@mac.com>
Reviewed-By: Michael Dawson <michael_dawson@ca.ibm.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
This commit add a target specific variable named 'build_type' to the
node and node_g targets.
The motivation for doing this is that both targets share the same
prerequisites, and the recepies are the same (apart from the build type)
so some duplication can be avoided.
PR-URL: https://github.com/nodejs/node/pull/33925
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
On Windows there is a program "find.exe" located in
C:\Windows\System32, which is usually in the PATH before
MSYS version of that program (required for running makefile).
The Windows version of the program uses different CLI syntax,
which results in errors like
"File not found - *node_modules*"
This commit specifies the full path to the program, which is also
properly handled by MSYS on Windows.
PR-URL: https://github.com/nodejs/node/pull/33136
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Prior to this commit, all <pre> tags were being
highlighted as JavaScript. This has been corrected
to syntax highlight all languages appearing in the
API reference docs. This was accomplished by using
highlight.js instead of SHJS for the frontend lib.
* remove SHJS JavaScript code
* add highlight.js bundle
* fix script tags to reflect replacement
* migrate CSS to use highlight.js classes
* add appropriate documentation
* ensure api_assets README.md stays interal
Fixes: https://github.com/nodejs/node/issues/33363
PR-URL: https://github.com/nodejs/node/pull/33442
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Ujjwal Sharma <ryzokuken@disroot.org>
* Updated cpp style guide file name and location and fixed links to
this file.
* Updated collaborator guide file name and location and fixed links
to this file.
* Updated documentation style guide file name and location and updated
links referencing the file.
* Moved files to appropriate location and updated naming style for
some of them.
Fixes: https://github.com/nodejs/node/issues/31741
PR-URL: https://github.com/nodejs/node/pull/31792
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Sam Roberts <vieuxtech@gmail.com>
Update Makefile so that only markdown files that have changed will be
linted. Currently, if one file in doc/api has changed, all files in
doc/api are linted.
On Windows, the lint-md task currently lints all files regardless of
whether any files has changed, and that behavior is unchanged here.
A further improvement is that when tools/lint-md.js is rebuilt, the
timestamp file is removed so that all files are linted again. This is
because rebuilding lint-md.js can introduce new rules or modify existing
rules, so re-linting everything helps make sure that accidental breakage
doesn't slip by unnoticed.
PR-URL: https://github.com/nodejs/node/pull/31923
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Co-authored-by: @patrickhousley
Fixes to Python tests to ensure that the following all pass:
1. __python2 -m pytest ./test ./tools__ # 30 tests pass
2. __python3 -m pytest ./test ./tools__ # 30 tests pass
3. __python2 -m unittest discover -s ./test/tools__ # 1 test passes
4. __python3 -m unittest discover -s ./test/tools__ # 1 test passes
5. __PYTHON=python2 make tooltest__ # 1 test passes
6. __PYTHON=python3 make tooltest__ # 1 test passes
This is a subset of #30033
PR-URL: https://github.com/nodejs/node/pull/30340
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: David Carlier <devnexen@gmail.com>
- Updated the tooltest target to run unittest module
- Renamed test/tools/test-js2c.py to be discoverable by unittest module
- Added test class for `configure` shell script
- Added a test to ensure `configure` script exits with status code zero
when passed the `--ninja` flag
Closes: https://github.com/nodejs/node/issues/29415
PR-URL: https://github.com/nodejs/node/pull/30033
Reviewed-By: Sam Roberts <vieuxtech@gmail.com>
Reviewed-By: Christian Clauss <cclauss@me.com>
Linter failures in the `test-doc` make target were not failing the
build if the subsequent `doctools` test passed as its exit code
wasn't being preserved.
Make the `lint` target a dependency of `test-doc` so that it is
outside of the `node_use_openssl` guard -- its own dependencies
have their own guards where necessary and the targets that don't
require an available node (e.g. the C++ linters) will be allowed
to run.
PR-URL: https://github.com/nodejs/node/pull/30012
Reviewed-By: Sam Roberts <vieuxtech@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: Daniel Bevenius <daniel.bevenius@gmail.com>
Reviewed-By: Gireesh Punathil <gpunathi@in.ibm.com>
Run cleanup for all relevant subdirectories, and exclude generated
source files (e.g. `out/Release/...`) from coverage reporting.
Also enable running `make cctest` again (It’s unclear to me why this
was disabled, as it does reduce coverage unnecessarily).
PR-URL: https://github.com/nodejs/node/pull/29487
Reviewed-By: Rich Trott <rtrott@gmail.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Benjamin Gruenbaum <benjamingr@gmail.com>
Reviewed-By: Jiawen Geng <technicalcute@gmail.com>
This commit removes the usage of the CI_DOC variable in the test-doc
recipe and specifies the doctool argument to tools/test.py explicitly.
The motivation for this is that the build is taking longer time and
this is mostly due to tests being run twice as the CI_DOC
variable will be empty in most cases (when not using --without-ssl).
This change was introduced with/after Commit
9039af83a3 ("build: skip test-ci doc
targets if no crypto") and while I though it might make sense to change
the setting of CI_DOC I not sure about the implications that might have
to our CI environment. It currently looks like this:
ifeq ($(node_use_openssl), false)
CI_DOC := doctool
else
CI_DOC =
endif
Which is setting CI_DOC to doctool if there is no crypto support which
not available. But perhaps this should be be the other way around,
changing the order or updating condition to be true.
PR-URL: https://github.com/nodejs/node/pull/29375
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Yongsheng Zhang <zyszys98@gmail.com>
Reviewed-By: David Carlier <devnexen@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
This commit moves the tooltest target to come before the jstest target
to make the console output consistent with it was before I introduced
the tooltest target.
Currently the output looks like this which is might give the impression
that only one test was run:
/Applications/Xcode.app/Contents/Developer/usr/bin/make -s tooltest
.
----------------------------------------------------------------------
Ran 1 test in 0.000s
Compared to the usual:
[03:58|% 100|+ 2739|- 0]: Done
PR-URL: https://github.com/nodejs/node/pull/29220
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: James M Snell <jasnell@gmail.com>
The description for the --link-module configuration option is as
follows:
$ ./configure --help | grep -A 5 'link-module'
--link-module=LINKED_MODULE
Path to a JS file to be bundled in the binary as a
builtin. This module will be referenced by path
without extension; e.g. /root/x/y.js will be
referenced via require('root/x/y'). Can be used
multiple times
This lead me to think that it was possible to specify a file like this:
$ ./configure --link-module=something.js
$ NODE_DEBUG=mkcodecache make -j8
This will lead to a compilation error as an entry in the source_ map in
node_javascript.cc will end up having an empty string as its key:
source_.emplace("", UnionBytes{_raw, 105});
This will then be used by CodeCacheBuilder when it iterates over the
module ids, which will lead to the following compilation errors:
/node/out/Release/obj/gen/node_code_cache.cc:12:23: warning:
ISO C++17 does not allow a decomposition group to be
empty [-Wempty-decomposition]
static const uint8_t [] = {
^
/node/out/Release/obj/gen/node_code_cache.cc:12:22: warning:
decomposition declarations are a C++17 extension [-Wc++17-extensions]
static const uint8_t [] = {
^~
/node/out/Release/obj/gen/node_code_cache.cc:12:1: error:
decomposition declaration cannot be declared 'static'
static const uint8_t [] = {
^~~~~~
/node/out/Release/obj/gen/node_code_cache.cc:12:22: error:
decomposition declaration cannot be declared with type 'const uint8_t'
(aka 'const unsigned char'); declared type must be 'auto' or
reference to 'auto'
static const uint8_t [] = {
^
/node/out/Release/obj/gen/node_code_cache.cc:12:22: error:
excess elements in scalar initializer
static const uint8_t [] = {
^
/node/out/Release/obj/gen/node_code_cache.cc:660:7: error:
expected expression
,
^
/node/out/Release/obj/gen/node_code_cache.cc:661:24: error:
no matching function for call to 'arraysize'
static_cast<int>(arraysize()), policy
^~~~~~~~~
../src/util.h:667:18: note: candidate function template not viable:
requires 1 argument, but 0 were provided
constexpr size_t arraysize(const T (&)[N]) {
^
2 warnings and 5 errors generated.
This commit suggests that passing a single file be allowed by modifying
tools/js2c.py.
PR-URL: https://github.com/nodejs/node/pull/28443
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
Reviewed-By: Rich Trott <rtrott@gmail.com>
Shelley Vohr