lib: drop Python 2 support in find-python.js #2333

DeeDeeG · 2021-02-15T22:22:22Z

Checklist

npm install && npm test passes
tests are included (I updated existing tests)
documentation is changed or added
commit message follows commit guidelines

Description of change

Follow-up to #2300 and #2318.

(Now that gyp-next no-longer supports Python 2, node-gyp should stop finding and trying to use Python 2 to run gyp-next with.)

This Pull Request:

Updates the accepted versions of Python in lib/find-python.js from '2.7.x || >=3.5.0' to just '>=3.6.0'
Removes checks attempting to find Python 2 (on all platforms)
Updates the fallback paths (default Python install paths) to check on Windows
Uses the -3 flag with the py.exe launcher (on Windows), so that it looks for/executes only a Python 3 executable
Updates the tests to reflect the above changes

Python 2 isn't supported in gyp-next anymore, so we should stop finding/trying to use Python 2.

Note: This commit is required for the previous commit to pass tests. I have left it separate for clarity during review of this PR. But it should be squashed with the previous commit if/when merging.

This reverts commit c255ffb. Note: My intent here is not to actually restore any Python 2 support, just to make it clearer where the comment text originally came from. It would be fair to squash this with the next commit if/when merging.

Ensures the py.exe launcher only finds Python 3. (Only applies to Windows users.)

lib/find-python.js

Co-authored-by: Christian Clauss <cclauss@me.com>

cclauss

LGTM! Thanks.

Note: This is required for the previous commit to pass tests. Please squash this with the previous commit if/when merging.

Paths are installed with a "-32" suffix: "Python[version]-32" I forgot to add the "-32" in previous commits. Note: It would be good to squash this with the previous two commits.

lib/find-python.js

32-bit Windows doesn't have a "Program Files (x86)" folder, only "Program Files".

Reduces boilerplate code. Also, refactored to drop the dependency on the 'path' module.

mfonville · 2021-02-19T08:11:01Z

lib/find-python.js

+const winDefaultLocationsArray = []
+
+for (const majorMinor of ['39', '38', '37', '36']) {
+  winDefaultLocationsArray.push(


Instead of fixed Users and Program Files shouldn't this be with the environment variables %ProgramFiles%, %ProgramFiles(x86)%, %ProgramW6432% and %LocalAppData%?

Because these folders can be set to be on other disks than the system drive, and avoids potential localization issues in the path.

@mfonville Thanks for the suggestion. I tried your suggestion on another branch: DeeDeeG@e57fdcc

It does work. It's also a bit more involved. After some quick research and testing, I am somewhat doubtful whether there are circumstances where the paths actually change based on localization or user customization. See my thoughts below. If you have examples or more information about when/if the paths are actually different, please let me know.

At the moment I am inclined to believe this is not necessary. But I would not mind being proved wrong. Thanks.

More thoughts on this (click to expand):

I wish I knew in what circumstances the "Program Files" or "AppData" folders are localized or moved, if there are any such circumstances. I do not think the paths are truly localized on-disk, nor movable, in Windows 10. It's unclear to me if they've ever been truly localized on-disk, or movable in previous Windows releases.

(Paths were slightly different in Windows XP, but Windows XP is not compatible with the latest node-gyp. Python 3.4 is the last version supporting Windows XP. node-gyp master branch now requires Python 3.6 and up, so no XP support anymore. And only Windows 10 is in "mainstream" support by Microsoft now.)

Regarding custom locations for Program Files or AppData: Is anyone here aware of a proper way to move these folders? (I am aware that folders such as Documents and Videos can be easily moved to another drive. But can "Program Files", user home folders, or "AppData" be moved off of the system drive? I have looked this up online, and the overall impression I am getting is that there is no official way to move these, and that moving them may break things.)

Regarding localization, or Windows in languages other than English: I don't see the paths being actually localized on the disk. (On Windows 7 or 10 in Spanish, I see it presented visually in the File Explorer as "C: > Usuarios > [user]", but the moment you try to edit the path in the address bar, it changes to C:\Users\[user]. The env var LOCALAPPDATA is still C:\Users\[user]\AppData\Local. The env var ProgramFiles is still C:\Program Files. Even while this shows as C: > Archivos de programa in File Explorer, the moment you try to edit it, it changes to C:\Program Files. I believe the "non-localized" versions are the true paths, and the localizations are superficial. At least this is the case in my testing on the latest Windows 10, and on an install made from an old Windows 7 iso.)

If XP is out of the picture, then I do think localization is probably indeed not an issue anymore. Windows 10 does this nicely.

But about custom paths, to be honest I am not working with Windows daily anymore (switched to Linux many years ago). But when a couple of years ago I did do some volunteering with managing some Windows systems, we could set custom locations for these paths through policies. And also I have seen customized paths when computers were upgraded from a HDD to a SSD, where these paths were customized to optimize performance.
I am not sure how Windows 10 handles it these days, but imho those environment variables are more reliable than any hardcoded path. So better to use the variables :-)

Thanks for that explanation. I also found this official Microsoft document about moved user folders: https://docs.microsoft.com/en-us/troubleshoot/windows-server/user-profiles-and-logon/relocation-of-users-and-programdata-directories

So apparently user directories can be moved to a custom location. (Maybe this applies only to Windows Server? I'd say node-gyp is not any less important for Windows Server users. It still seems like a bit of an obscure feature and is discouraged in the docs I linked.)

I'm a little more inclined to do this (DeeDeeG@e57fdcc) now. Truth be told, it's not much more complex. Given even a mild justification for end-user benefit, I'd want to do it.

NOTE TO MAINTAINERS/REVIEWERS: Please let me know if you'd like me to add this (DeeDeeG@e57fdcc) to the PR. Thank you.

And most of all, there is no downside to the change I think.

Go for it. This code is easy to read/ understand.

I know that is the Swiss in me but...

Please sort your const declarations in the order that they are used to improve readability.

More research: We definitely don't have to worry about Windows XP or Vista.

Node v6 refuses to run on anything older than Windows 7. Node v5 is the newest Node that will run on Windows Vista. Node v5 cannot even run the latest node-gyp, as node-gyp user newer JavaScript features that aren't supported in Node 5. In conclusion: modern node-gyp won't run at all on Windows XP or Vista, regardless of this PR.

environment variables [. . .] %ProgramW6432% [. . .]

@mfonville thanks for mentioning that. I believe I will need to switch from process.env.ProgramFiles to process.env.ProgramW6432 to get the "not (x86)" Program Files.

(Note: this is important when running 32-bit Node on 64-bit Windows, which is common when building 32-bit code on a 64-bit host, such as in CI.)

It appears we have to use process.env.ProgramW6432 to get the not-architecture-specific Program Files location. From 32-bit Node on a 64-bit host, process.env.ProgramFiles returns [driveLetter]:\Program Files (x86). As such, ProgramW6432 is the only easy way to get [driveLetter]:\Program Files from the environment across all combinations of host arch/node arch.

(Official documentation says that the env var ProgramW6432 only exists as of Windows 7 or newer, but that's okay. node-gyp doesn't even run on Windows Vista or older.)

See:

https://docs.microsoft.com/en-us/windows/win32/winprog64/wow64-implementation-details?redirectedfrom=MSDN#environment-variables

https://serverfault.com/questions/413320/programfiles-differences-on-64bit-windows

Okay, I committed this in the PR now. Program Files, Program Files (x86) and AppData\Local are set from environment variables now.

test/test-find-python.js

DeeDeeG · 2021-02-21T07:19:44Z

lib/find-python.js

 const log = require('npmlog')
 const semver = require('semver')
 const cp = require('child_process')
 const extend = require('util')._extend // eslint-disable-line
 const win = process.platform === 'win32'
 const logWithPrefix = require('./util').logWithPrefix

+const systemDrive = process.env.SystemDrive || 'C:'
+const username = process.env.USERNAME || process.env.USER


I think this should fall back to assigning some string to username. Even something not very useful like '' or 'User'.

(It's possible both process.env.USERNAME and process.env.USER are undefined. Having a not very useful path to check is better than seeing Uncaught ReferenceError: username is not defined.)

I actually like the error message better than unsuccessfully trying to sweep the problem under the carpet. The message makes it clear what the problem is.

Please do not worry about the Python 4 case. We will cross that bridge later.

Thinking a bit more...

If username is undefined and we can find a globally installed acceptable Python then we should use that Python without complaint. If we fail to find an acceptable Python and username is undefined then we should raise the error.

You can also get the username via os.userInfo().

Thanks for that suggestion @cjihrig. os.userInfo() should always return a username string. I committed that as a fallback. I think this can reasonably be expected to prevent Uncaught ReferenceError: username is not defined.

Note: There was apparently no risk of Uncaught ReferenceError after all.

If you write .someNotDefinedThing off of an object or array, you get undefined as the result, but no ReferenceError. Since process.env returns an object, doing process.env.THIS_IS_NOT_DEFINED does not produce a ReferenceError.

For example:

console.log(`Here is my undefined property: ${{}.some_undefined_property}`)

Outputs the string:

Here is my undefined property: undefined

So we would construct a Python path resembling this: C:\\Users\\undefined\\AppData\\Local\\Programs\\Python\\Python39\\python.exe

(Note the Users\\undefined bit.)

I'm happy with the os.userInfo() fallback, regardless. It makes it pretty much guaranteed we can find the username. This is just an interesting finding from my tests.

Related to this, still an issue when the username is not defined, only the user ID e.g. that is common in 'non-root' containers for OpenShift https://docs.openshift.com/container-platform/3.3/creating_images/guidelines.html#openshift-container-platform-specific-guidelines.

I have no name!@52545ce360b8$ node-gyp --help os.js:272 throw new ERR_SYSTEM_ERROR(ctx); ^ SystemError [ERR_SYSTEM_ERROR]: A system error occurred: uv_os_get_passwd returned ENOENT (no such file or directory) at Object.userInfo (os.js:272:11) at Object.<anonymous> (/opt/bitnami/node/lib/node_modules/node-gyp/lib/find-python.js:11:76) at Module._compile (internal/modules/cjs/loader.js:999:30) at Object.Module._extensions..js (internal/modules/cjs/loader.js:1027:10) at Module.load (internal/modules/cjs/loader.js:863:32) at Function.Module._load (internal/modules/cjs/loader.js:708:14) at Module.require (internal/modules/cjs/loader.js:887:19) at require (internal/modules/cjs/helpers.js:74:18) at Object.<anonymous> (/opt/bitnami/node/lib/node_modules/node-gyp/lib/configure.js:11:18) at Module._compile (internal/modules/cjs/loader.js:999:30) { code: 'ERR_SYSTEM_ERROR', info: { errno: -2, code: 'ENOENT', message: 'no such file or directory', syscall: 'uv_os_get_passwd' }, errno: [Getter/Setter], syscall: [Getter/Setter] }

Python is in the path so a good option could be to add a fallback or not return an error for the line

const username = process.env.USERNAME || process.env.USER || require('os').userInfo().username

These paths may be localized to other languages, or placed on another drive. (These paths appear to be mostly standardized on Windows 10, but they can apparently be relocated on Windows Server, and may be localized on some older version(s) of Windows.)

This should be a reliable fallback in case the environment variables have been deleted or unset.

Note: This fixes an oversight in the previous commit, and is required for the previous commit to pass tests. Please squash this with the previous commit if/when merging.

DeeDeeG · 2021-02-22T20:06:05Z

I think the py.exe launcher works really well. I'd like to suggest it be tried before all these winDefaultLocations entries. I'm not sure if that's out-of-scope for this PR though.

(I noticed while testing this pull request that the py.exe launcher is run after checking all these potential paths from winDefaultLocations.)

(Again, not sure if it would be part of this PR or not, but here's an example implementation: DeeDeeG@c39ad57).

cclauss · 2021-02-22T20:20:17Z

@cjihrig @mfonville Are you folks satisfied with the current state of this PR? There is nice work here and it would be good to have more PR approvals besides @gengjiawen and myself.

mfonville · 2021-02-22T20:24:39Z

I am happy with its current state. You could still consider @DeeDeeG's last suggestion to try py.exe before all the generated entries. That would probably be the logical thing to do.

cclauss · 2021-02-22T21:02:01Z

@mfonville

https://github.com/nodejs/node-gyp/pull/2333/files
Review changes
Approve
Submit review

DeeDeeG · 2021-03-20T04:44:08Z

Reminder: I think it might be neat to prioritize trying the py.exe launcher, before checking all the default install paths, on Windows.

I have an implementation of this ready to go. (DeeDeeG@c39ad57) Please let me know if this is preferred. And/or if you want that, but as another PR.

rvagg · 2021-03-22T04:50:32Z

I'm happy to include this in v8.0.0 but there's been some additional commits since @gengjiawen and @cclauss approved, would either of you mind taking a look over the current state?
I'll squash it into a single commit for merging, unless anyone thinks there's separate things going on here worth individuating.

gengjiawen · 2021-03-26T09:31:00Z

@cclauss Still looks good to you ?

cclauss · 2021-03-26T10:58:23Z

This PR already has my approval.

* gyp: update gyp to 0.2.1 PR-URL: nodejs/node-gyp#2092 Reviewed-By: Rod Vagg <rod@vagg.org> * deps: replace mkdirp with {recursive} mkdir only supported on Node.js 10+ Closes: #2084 PR-URL: nodejs/node-gyp#2123 Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * doc: update acid test and introduce curl|bash test script PR-URL: nodejs/node-gyp#2105 Reviewed-By: Rod Vagg <rod@vagg.org> * doc: update catalina xcode clt download link PR-URL: nodejs/node-gyp#2133 Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Christian Clauss <cclauss@me.com> * v7.0.0: bump version and update changelog PR-URL: nodejs/node-gyp#2124 * deps: increase "engines" to "node" : ">= 10.12.0" Makes npm warn users if they are using an unsupported Node version. Refs: nodejs/node-gyp#2123 PR-URL: nodejs/node-gyp#2153 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> Reviewed-By: Rod Vagg <rod@vagg.org> * doc: silence curl for macOS Catalina acid test PR-URL: nodejs/node-gyp#2150 Reviewed-By: Rod Vagg <rod@vagg.org> * docs: note that node-gyp@7 should solve Catalina CLT issues PR-URL: nodejs/node-gyp#2156 Reviewed-By: Christian Clauss <cclauss@me.com> * build: support apple silicon (arm64 darwin) builds Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> PR-URL: nodejs/node-gyp#2165 * gyp: update gyp to 0.4.0 Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> PR-URL: nodejs/node-gyp#2165 * build: add update-gyp script Co-authored-by: Christian Clauss <cclauss@me.com> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> Reviewed-By: Christian Clauss <cclauss@me.com> PR-URL: nodejs/node-gyp#2167 * v7.1.0: bump version and update changelog * doc: drop the --production flag for installing windows-build-tools This isn't needed, and was probably copy-pasted from windows-build-tools' README.md, which has since been changed to drop the `--production` flag from the install instructions. PR-URL: nodejs/node-gyp#2206 Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Rod Vagg <rod@vagg.org> * ci: switch to GitHub Actions Co-authored-by: Christian Clauss <cclauss@me.com> Co-authored-by: Matias Lopez <imatlopez@gmail.com> PR-URL: nodejs/node-gyp#2210 Closes: #2127 Closes: #2209 * doc: replace status badges with new Actions badge PR-URL: nodejs/node-gyp#2218 Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Matias Lopez <imatlopez@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> * test: GitHub Actions: Test on Python 3.9 From python: [3.6, 3.7, 3.8] --> python: [3.6, 3.8, 3.9] because if things work on Python 3.6 and 3.8 then they should work on 3.7. https://www.python.org/downloads/release/python-390/ PR-URL: nodejs/node-gyp#2230 Reviewed-By: Shelley Vohr <shelley.vohr@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> * lib: better log message when ps fails PR-URL: nodejs/node-gyp#2229 Reviewed-By: Bartosz Sosnowski <bartosz@janeasystems.com> Reviewed-By: Rod Vagg <rod@vagg.org> * gyp: update gyp to 0.6.1 Closes: nodejs/node-gyp#2236 PR-URL: nodejs/node-gyp#2238 Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Myles Borins <myles.borins@gmail.com> * deps: update deps to match npm@7 PR-URL: nodejs/node-gyp#2240 Reviewed-By: Richard Lau <riclau@uk.ibm.com> * v7.1.1: bump version and update changelog PR-URL: nodejs/node-gyp#2239 Reviewed-By: Christian Clauss <cclauss@me.com> * doc: add cmd to reset `xcode-select` to initial state PR-URL: nodejs/node-gyp#2235 Reviewed-By: Christian Clauss <cclauss@me.com> * gyp: update gyp to 0.6.2 Refs: https://github.com/nodejs/gyp-next/releases/tag/v0.6.2 PR-URL: nodejs/node-gyp#2241 Reviewed-By: Rod Vagg <rod@vagg.org> * v7.1.1: bump version and update changelog PR-URL: nodejs/node-gyp#2242 * doc: add missing `sudo` to Catalina doc PR-URL: nodejs/node-gyp#2244 Reviewed-By: Rod Vagg <rod@vagg.org> * ci: migrate deprecated grammar (#2285) PR-URL: nodejs/node-gyp#2285 Reviewed-By: Richard Lau <rlau@redhat.com> * doc: updated README.md to copy easily (#2281) PR-URL: nodejs/node-gyp#2281 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * gyp: update gyp to v0.7.0 (#2284) PR-URL: nodejs/node-gyp#2284 Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Christian Clauss <cclauss@me.com> * doc: update macOS_Catalina.md (#2293) PR-URL: nodejs/node-gyp#2293 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * gyp: update gyp to v0.8.0 (#2318) PR-URL: nodejs/node-gyp#2318 Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * lib: avoid changing process.config (#2322) PR-URL: nodejs/node-gyp#2322 Refs: nodejs/node#36902 Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Rod Vagg <rod@vagg.org> * gyp: remove support for Python 2 (#2300) PR-URL: nodejs/node-gyp#2300 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * lib: migrate requests to fetch (#2220) PR-URL: nodejs/node-gyp#2220 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * lib: drop Python 2 support in find-python.js (#2333) Co-authored-by: Christian Clauss <cclauss@me.com> PR-URL: nodejs/node-gyp#2333 Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * ci: update actions/setup-node to v2 (#2302) PR-URL: nodejs/node-gyp#2302 Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Richard Lau <rlau@redhat.com> Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * doc: add downloads badge (#2352) PR-URL: nodejs/node-gyp#2352 Reviewed-By: Christian Clauss <cclauss@me.com> Reviewed-By: Richard Lau <rlau@redhat.com> * deps: sync mutual dependencies with npm Sync with npm 7.7.0 PR-URL: nodejs/node-gyp#2348 Reviewed-By: Rod Vagg <rod@vagg.org> Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * gyp: Improve our flake8 linting tests PR-URL: nodejs/node-gyp#2356 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> * gyp: update gyp to v0.8.1 (#2355) PR-URL: nodejs/node-gyp#2355 Reviewed-By: Jiawen Geng <technicalcute@gmail.com> Reviewed-By: Christian Clauss <cclauss@me.com> * v8.0.0: bump version and update changelog * doc: fix v8.0.0 release date PR-URL: nodejs/node-gyp#2346 * meta: add `release-please-action` for automated releases (#2395) Co-authored-by: gengjiawen <technicalcute@gmail.com> * lib: fail gracefully if we can't find the username (#2375) * lib: log as yes/no whether build dir was created (#2370) This bit of logging apparently expected to be given a boolean, but was receiving either a path or undefined based on the result of fs.mkdir. Now it prints either "Yes" or "No", rather than printing either a path or "undefined", respectively. * doc: Update README.md Visual Studio Community page polski to auto (#2371) changed URL of Visual Studio Community from a default polski URL to the one without the lenguage code * doc: remove redundant version info (#2403) * feat(gyp): update gyp to v0.9.1 (#2402) * chore: release 8.1.0 (#2418) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * lib: deep-copy process.config during configure (#2368) * fix: change default gyp update message (#2420) * fix: add error arg back into catch block for older Node.js users * chore: fix typos discovered by codespell (#2442) * Initial Commit * doc(wiki): Created "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): fixed node-serialport link * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): add bcrypt * doc(wiki): change bcrypt url to binding.gyp file * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): add one more example * doc(wiki): Add a link to the node-midi binding.gyp file. * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Adds node-inotify and v8-profiler * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Adds node-fann * doc(wiki): Add node-canvas * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Created Linking to OpenSSL (markdown) * doc(wiki): Updated Home (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated Linking to OpenSSL (markdown) * doc(wiki): added levelup * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Created Updating npm's bundled node gyp (markdown) * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): Created Visual Studio 2010 Setup (markdown) * doc(wiki): Updated Home (markdown) * doc(wiki): Created Common issues (markdown) * doc(wiki): Updated Home (markdown) * doc(wiki): Add helpful information * doc(wiki): Created Error: "pre" versions of node cannot be installed (markdown) * doc(wiki): Updated Error: "pre" versions of node cannot be installed (markdown) * doc(wiki): Updated Home (markdown) * doc(wiki): fix link to gyp file used to build libsqlite3 * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Bumping Python version from 2.3 to 2.7 as per the node-gyp readme * doc(wiki): Add node-openvg-canvas and node-openvg. * doc(wiki): Updated Home (markdown) * doc(wiki): Adding link to node-cryptopp's gyp file * doc(wiki): Updated Linking to OpenSSL (markdown) * doc(wiki): add topcube, node-osmium, and node-osrm * doc(wiki): Created use of undeclared identifier 'TypedArray' (markdown) * doc(wiki): Created Visual studio 2012 setup (markdown) * doc(wiki): Destroyed Visual studio 2012 setup (markdown) * doc(wiki): Correcting the link to node-osmium * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Fix link to node-zipfile * doc(wiki): Explicit link to Visual C++ 2010 Express * doc(wiki): Added tip about resolving frustrating LNK1181 error * doc(wiki): Updated node-levelup to node-leveldown (broken links) * doc(wiki): Added details for properly fixing * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Added nk-mysql (nodamysql) * doc(wiki): Added nk-xrm-installer .gyp references, including .py scripts for providing complete reference to examples of fetching source via http, extracting, and moving files (as opposed to copying) * doc(wiki): Note: VS2010 seems to be no longer available! VS2013 or nothing! * doc(wiki): node-sass in the wild * doc(wiki): Clarification + direct link to VS2010 * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): if ouns that the -h did not help. I founs on github that there was support for visual studio 2015, while i couldn't install node-red beacuse it kept telling me the key 2015 was missing. looking in he gyp python code i found the local file was bot up t dat with the github repo. updating took several efforts before i tried to drop the -g option. * doc(wiki): sorry, forgot to mention a specific windows version. * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Added Ghostscript4JS * doc(wiki): I highly missing it in common issue as every windows biggner face that issue * doc(wiki): ADDED: Node.js binding to OpenCV * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Adding the sharp library to the list * doc(wiki): node-srs was a 404 * doc(wiki): C++ build tools version upgraded * doc(wiki): Destroyed Visual Studio 2010 Setup (markdown) * doc(wiki): Updated Home (markdown) * doc(wiki): Lower case L * doc(wiki): Updated "binding.gyp" files out in the wild (markdown) * doc(wiki): Make changes discussed in nodejs/node-gyp#2416 * doc(wiki): Drop in favor of * doc(wiki): Different commands for Windows npm v6 vs. v7 * doc(wiki): Improve Unix instructions * doc(wiki): Updated Updating npm's bundled node gyp (markdown) * doc(wiki): If permissions error, please try and then the command. * doc(wiki): move wiki docs into doc/ * doc(wiki): link to docs/ from README * doc(wiki): safer doc names, remove unnecessary TypedArray doc * ci: GitHub Actions Test on node: [12.x, 14.x, 16.x] (#2439) * Add title to node-gyp version document (#2452) * Add title to node-gyp version document * Update Updating-npm-bundled-node-gyp.md * fix: doc how to update node-gyp independently from npm * fix: missing spaces * ISSUE_TEMPLATE.md: Instructions for old versions (#2470) * ISSUE_TEMPLATE.md: Instructions for old versions Also, add a caution about `node sass` being deprecated. * Update .github/ISSUE_TEMPLATE.md Co-authored-by: Rod Vagg <rod@vagg.org> Co-authored-by: Rod Vagg <rod@vagg.org> * chore(deps): bump tar from 6.1.0 to 6.1.2 (#2474) Addresses GHSA-3jfq-g458-7qm9 and GHSA-r628-mhmh-qjhw * doc: correct link to "binding.gyp files out in the wild" (#2483) correct link to "binding.gyp files out in the wild" * feat(gyp): update gyp to v0.9.6 (#2481) * chore: release 8.2.0 * chore: refactor the creation of config.gypi file * test: Python 3.10 was release on Oct. 4th (#2504) * chore(deps): bump make-fetch-happen from 8.0.14 to 9.1.0 The breaking change in this module was a cache parameter that `node-gyp` is not using, so this module is not affected. * feat(gyp): update gyp to v0.10.0 (#2521) * chore: release 8.3.0 * feat: support vs2022 (#2533) * feat: build with config.gypi from node headers * chore: release 8.4.0 * docs: fix typo in powershell node-gyp update * deps: npmlog@6.0.0 * fix: windows command missing space (#2553) * chore: release 8.4.1 * chore: add minimal SECURITY.md (#2560) * doc: Rename and update Common-issues.md --> docs/README.md (#2567) Update the common problems to track with current issues on this repo and shorten the URL to just https://github.com/nodejs/node-gyp/tree/master/docs * docs: title match content (#2574) * docs: Add notes/disclaimers for upgrading the copy of node-gyp that npm uses (#2585) * docs: rephrase explanation of which node-gyp is used by npm (#2587) * doc: Update Python versions (#2571) * Add Python 3.10 * Drop Python 3.6 which [EOLs on 23 Dec. 2021](https://devguide.python.org/#status-of-python-branches) * macOS: clarify `Xcode Command Line Tools` standalone vs. from full Xcode * Window: Use the same URL as https://github.com/nodejs/node/blob/master/BUILDING.md#windows * deps!: increase "engines" to "node" : "^12.22 || ^14.13 || >=16" (#2601) Makes npm warn users if they are using an unsupported Node version. * deps: make-fetch-happen@10.0.1 The breaking change was dropping node10 support, which node-gyp has already done. * fix: update make-fetch-happen to a minimum of 10.0.3 * added node-heapdump binding.gyp * fix: _ in npm_config_ env variables * lib: add lib.target as path for searching libnode on z/OS * chore: release 9.0.0 * test: Upgrade GitHub Actions (#2623) * doc: update docs/README.md with latest version number * fix: typo on readme * fix: new ca & server certs, bundle in .js file and unpack for testing bundling in certs.js rather than including the raw files should avoid some false positives that low-quality security scanners keep on complaining about. * fix: extend tap timeout length to allow for slow CI * Add Python symlink to path (for non-Windows OSes only) (#2362) * lib: create a Python symlink and add it to PATH Helps to ensure a version of Python validated by lib/find-python.js is used to run various Python scripts generated by gyp. Known to affect gyp-mac-tool, probably affects gyp-flock-tool as well. These Python scripts (such as `gyp-mac-tool`) are invoked directly, via the generated Makefile, so their shebang lines determine which Python binary is used to run them. The shebang lines of these scripts are all `#!/usr/bin/env python3`, so the first `python3` on the user's PATH will be used. By adding a symlink to the Python binary validated by find-python.js, and putting this symlink first on the PATH, we can ensure we use a compatible version of Python to run these scripts. (Only on Unix/Unix-like OSes. Symlinks are tricky on Windows, and Python isn't used at build-time anyhow on Windows, so this intervention isn't useful or necessary on Windows. A similar technique for Windows, no symlinks required, would be to make batch scripts which execute the target binary, much like what Node does for its bundled copy of npm on Windows.) * test: update mocked graceful-fs for configure test Add missing functions "unlink()" and "symlink()" to mocked module. * lib: log any errors when creating Python symlink Warn users about errors, but continue on in case the user does happen to have new enough Python on their PATH. (The symlinks are only meant to fix an issue in a corner case, where the user told `node-gyp` where new enough Python is, but it's not the first `python3` on their PATH. We should not introduce a new potential failure mode to all users when fixing this bug. So no hard errors during the symlink process.) * lib: improve error formatting for Python symlink Logging the entire error object shows the stack twice, and all the other information is contained in the stack. It also messes with the order of what is logged. Rather than logging a bunch of redundant information in a messy way, we can log only the stack. Logging it in a separate log.warn() also gets rid of an extra space character at the beginning of the line. * lib: restore err.errno to logs for symlink errors This info (err.errno) is the only piece of information in the error object that is not redundant to err.stack. * lib: use log.verbose, not log.warn These messages aren't important enough to be `log.warn`s. Log as verbose only; they will also appear in full error output. * Clarify wording to redirect to macOS_Catalina.md (#2588) * build: update due to rename of primary branch * Migrate macOS acid test from master to main (#2686) Follow-on to #2495 * feat: Update function getSDK() to support Windows 11 SDK (#2565) * test: Upgrade GitHub Actions (#2701) * test: Upgrade GitHub Actions * node: 18x --> 18.x * test: Try msvs-version: [2016, 2019, 2022] (#2700) * test: Try msvs-version: [2016, 2019, 2022] * main, not master * Don't npm audit fix --force * fix: re-label (#2689) * chore: release 9.1.0 * lib: enable support for zoslib on z/OS (#2600) Check if zos-base.h is in the directory identified by environment variable ZOSLIB_INCLUDES if set; otherwise search for it from a set of candidates under nodeRootDir. Then pass it as -Dzoslib_include_dir=<path-found> to gyp_main.py for use in common.gypi to set 'includes_dir' when compiling addons. Co-authored-by: Gaby Baghdadi <baghdadi@ca.ibm.com> Co-authored-by: Gaby Baghdadi <baghdadi@ca.ibm.com> * chore: update dependency - nopt@6.0.0 (#2707) No functional changes, just dropping old node versions from engines, linting, and fixing CI. * fix: node.js debugger adds stderr (but exit code is 0) -> shouldn't throw (#2719) * fix: node.js debugger adds stderr (but exit code is 0) -> shouldn't throw * input.py: subprocess.Popen() -> subprocess.run() * feat(gyp): update gyp to v0.13.0 * feat: Add proper support for IBM i Python 3.9 on IBM i now properly returns "os400" for sys.platform instead of claiming to be AIX as it did previously. While the IBM i PASE environment is compatible with AIX, it is a subset and has numerous differences which makes it beneficial to distinguish, however this means that it now needs explicit support here. * Adding tarfile member sanitization to extractall() (#2741) Co-authored-by: TrellixVulnTeam <kasimir.schulz@trellix.com> * chore: release 9.2.0 (#2735) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * feat: support IBM Open XL C/C++ on z/OS (#2743) * feat: remove support for VS2015 in Node.js >=19 (#2746) * feat(gyp): update gyp to v0.14.0 (#2749) * chore: release 9.3.0 * Add Python 3.11 to the testing https://docs.python.org/3/whatsnew/3.11.html * ci: update python test matrix (#2774) * ci: drop python 3.6 from test matrix * Update .github/workflows/tests.yml Co-authored-by: Christian Clauss <cclauss@me.com> Co-authored-by: Christian Clauss <cclauss@me.com> * fix: increase node 12 support to ^12.13 (#2771) * chore: release 9.3.1 * feat: Upgrade Python linting from flake8 to ruff (#2815) [Ruff](https://beta.ruff.rs/) supports [over 500 lint rules](https://beta.ruff.rs/docs/rules) including bandit, isort, pylint, pyupgrade, and flake8 plus its plugins and is written in Rust for speed. This GitHub Action will provide contributors with intuitive GitHub Annotations. ![image](https://user-images.githubusercontent.com/3709715/223758136-afc386d2-70aa-4eff-953a-2c2d82ceea23.png) The `Required` in the checks below should be: 1. Removed from `flake8-annotation` and added to `ruff-annotation` which replaces it. 2. Removed from `isort` and added to `ruff` which replaces it. * Improved advise on repacing node-sass with sass (#2758) * Improved advise on repacing node-sass with sass * Update README.md * chore: get update-gyp.py to work with Python >= v3.5 (#2826) * chore: get update-gyp.py to work with Python v3.9 * Ruff ignore rule PLC1901 --------- Co-authored-by: Christian Clauss <cclauss@me.com> * doc: Update README.md (#2822) Co-authored-by: Christian Clauss <cclauss@me.com> * update make-fetch-happen to 11.0.3 (#2796) http-cache-semantics 4.1.0 is vulnerable https://www.cve.org/CVERecord?id=CVE-2022-25881 * docs: docs/README.md add advise about deprecated node-sass (#2828) * feat: add support for native windows arm64 build tools Visual Studio 2022 17.4 adds a native C++ compiler for Windows on ARM. This allows arm64 devices to leverage native build tools, leading to a 35% (or more) speed increase. https://devblogs.microsoft.com/visualstudio/arm64-visual-studio-is-officially-here/ Signed-off-by: Dennis Ameling <dennis@dennisameling.com> * fix: extract tarball to temp directory on Windows (#2846) * fix: check for errors while extracting downloaded tarball Signed-off-by: David Sanders <dsanders11@ucsbalum.com> * test: parallel installs Signed-off-by: David Sanders <dsanders11@ucsbalum.com> * fix: extract tarball to temp directory on Windows Signed-off-by: David Sanders <dsanders11@ucsbalum.com> --------- Signed-off-by: David Sanders <dsanders11@ucsbalum.com> * Migration from tap to mocha (#2851) * migrate from tap to mocha After make-fetch-happen update GitHub Actions started failing. Migrating from tap to mocha testing framework for GitHub Action stability. * write custom test reporter for more verbose output Implemented a simple custom mocha test reporter to replace the default one. Made test report more developer friendly. * fix: log statement is for devDir not nodedir (#2840) Signed-off-by: David Sanders <dsanders11@ucsbalum.com> * win,install: only download target_arch node.lib (#2857) Instead of downloading node.lib for all architectures, just download the one that will be needed. Install.js changed to enable downloading just node.lib for node versions that already have tarball downloaded and extracted. Not fetching lib now fails the installation. Increased installVersion because of the changes. Refs: nodejs/node-gyp#2847 * test: remove deprecated Node.js and Python (#2868) * test: remove deprecated node.js and python Removed Node.js v14.x and Python v3.7. Also added Node.js v20.x. * Update .github/workflows/tests.yml Co-authored-by: Christian Clauss <cclauss@me.com> --------- Co-authored-by: Christian Clauss <cclauss@me.com> * chore: release 9.4.0 * Sync deps and engines with npm (#2770) * feat!: update `engines.node` to `^14.17.0 || ^16.13.0 || >=18.0.0` * deps: nopt@^7.0.0 * feat: replace npmlog with proc-log * deps: standard@17.0.0 and fix linting errors * deps: which@3.0.0 - this also promiisifies the build command * deps: glob@8.0.3 * feat: drop rimraf dependency * fix: use fs/promises in favor of fs.promises * lib: find python checks order changed on windows (#2872) These changes favor py launcher over other checks excluding command line or npm configuration and environment variable checks. Also, updated supported python versions list. Fixes: nodejs/node-gyp#2871 * Fix reading msvs version on Windows (#2644) * fix: fix reading msvs version on windows * docs: Update windows installation instructions in README.md (#2882) * Update windows installation instructions in README.md * Fix Python lint error by using an f-string (#2886) * test: increase mocha timeout (#2887) * fix: create Python symlink only during builds, and clean it up after (#2721) * fix: create Python symlink only during builds, and clean it up after Previously in b9ddcd5 this was created during configuration, and the symlink persisted indefinitely. This causes problems with many tools that do not expect a codebase to include symlinks to external absolute paths. This PR largely reverts that commit, and instead writes the path to link to into the config, and then creates the symlink only temporarily during the build process, always deleting it afterwards. * assert install_path == self.output, f"{install_path} != {self.output}" --------- Co-authored-by: Christian Clauss <cclauss@me.com> * docs: README.md Do not hardcode the supported versions of Python (#2880) * Fix incorrect Xcode casing in README (#2896) * test: update expired certs (#2908) * doc: Add note about Python symlinks (PR 2362) to CHANGELOG.md for 9.1.0 (#2783) The PR for this change was merged without a prefixed name, such as "lib:" or "fix:". That means release-please didn't include it in the changelog for v9.1.0. This change did end up affecting users, though. (See issue 2713 and PR 2721). Therefore, I believe it should be noted in the CHANGELOG.md, so users can better understand the behavior they are seeing. * Python lint: ruff --format is now --output-format Fixes the failing `ruff` linting in GitHub Actions. * chore: GitHub Workflows security hardening (#2740) * build: harden tests.yml permissions Signed-off-by: Alex <aleksandrosansan@gmail.com> * build: harden release-please.yml permissions Signed-off-by: Alex <aleksandrosansan@gmail.com> * build: harden visual-studio.yml permissions Signed-off-by: Alex <aleksandrosansan@gmail.com> * Update release-please.yml --------- Signed-off-by: Alex <aleksandrosansan@gmail.com> * chore: empty commit to add changelog entries from #2770 feat!: update engines.node to ^14.17.0 || ^16.13.0 || >=18.0.0 deps: nopt@^7.0.0 feat: replace npmlog with proc-log deps: standard@17.0.0 and fix linting errors deps: which@3.0.0 fix: promisify build command deps: glob@8.0.3 feat: drop rimraf dependency fix: use fs/promises in favor of fs.promises * docs: update applicable GitHub links from master to main (#2843) Signed-off-by: David Sanders <dsanders11@ucsbalum.com> * feat(gyp): update gyp to v0.16.1 (#2923) * feat(gyp): update gyp to v0.15.1 * Add Python 3.12 to tests * Try to fix CI * Try specifying msvs-version * Modify the visual-studio matrix * Fix pythonLocation var * Fix Python tests * Get path * polish * feat(gyp): update gyp to v0.16.0 * feat(gyp): update gyp to v0.16.1 * CI: Don't install Python 'packaging' module (vendored in 'gyp-next' now) * Apply suggestions from code review * Upgrade to actions/checkout@v4 --------- Co-authored-by: Raymond Zhao <7199958+rzhao271@users.noreply.github.com> Co-authored-by: Christian Clauss <cclauss@me.com> * chore: add check engines script to CI (#2922) * deps: glob@10.3.10 (#2926) * deps: make-fetch-happen@13.0.0 (#2927) * deps: which@4.0.0 (#2928) * feat!: drop node 14 support (#2929) BREAKING CHANGE: `node-gyp` now supports node `^16.14.0 || >=18.0.0` * feat: convert all internal functions to async/await BREAKING CHANGE: All internal functions have been coverted to return promises and no longer accept callbacks. This is not a breaking change for users but may be breaking to consumers of `node-gyp` if you are requiring internal functions directly. * feat: convert internal classes from util.inherits to classes BREAKING CHANGE: the `Gyp` class exported is now created using ECMAScript classes and therefore might have small differences to classes that were previously created with `util.inherits`. * chore: misc testing fixes (#2930) * chore: misc test fixes * Sort test runs by os first * Use cross-env for test env var * Try sorting matrix params * Make FAST_TEST the default and rename to FULL_TEST * Separate helper functions to not need to export test obj in files * feat!: use .npmignore file to limit which files are published (#2921) * feat!: use package.json files to limit which files are published Fixes: #2372 * Use npmignore instead of package.json#files * Add update-gyp.py to npmignore * Add install to pack test * Use output var for pack dir * Move existing .gitignore entries to .npmignore * Sort git and npm ignores * Update and cleanup workflows * chore: run tests after release please PR * chore: release 10.0.0 (#2920) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> * chore: add parallel test logging * chore: lint fixes * chore: use platform specific timeouts in tests * fix: use local `util` for `findAccessibleSync()` The `findAccessibleSync()` function is in the local `util` module instead of Node.js' builtin `util` module. * chore: release 10.0.1 * Fix wrong changes that were on repo before merge * fix: output path generation for iOS --------- Signed-off-by: Dennis Ameling <dennis@dennisameling.com> Signed-off-by: David Sanders <dsanders11@ucsbalum.com> Signed-off-by: Alex <aleksandrosansan@gmail.com> Co-authored-by: Ujjwal Sharma <ryzokuken@disroot.org> Co-authored-by: Rod Vagg <rod@vagg.org> Co-authored-by: Dario Vladovic <d.vladimyr@gmail.com> Co-authored-by: DeeDeeG <DeeDeeG@users.noreply.github.com> Co-authored-by: Chia Wei Ong <ongchiawei@gmail.com> Co-authored-by: Samuel Attard <samuel.r.attard@gmail.com> Co-authored-by: Christian Clauss <cclauss@me.com> Co-authored-by: Shelley Vohr <shelley.vohr@gmail.com> Co-authored-by: Matias Lopez <imatlopez@gmail.com> Co-authored-by: Martin Midtgaard <martin.midtgaard@gmail.com> Co-authored-by: Valera Rozuvan <valera.rozuvan@gmail.com> Co-authored-by: Myles Borins <mylesborins@github.com> Co-authored-by: Karl Horky <karl.horky@gmail.com> Co-authored-by: Jiawen Geng <3759816+gengjiawen@users.noreply.github.com> Co-authored-by: மனோஜ்குமார் பழனிச்சாமி <smartmanoj42857@gmail.com> Co-authored-by: iMrLopez <8272737+iMrLopez@users.noreply.github.com> Co-authored-by: Michaël Zasso <targos@protonmail.com> Co-authored-by: Matias Lopez <imatlopez@users.noreply.github.com> Co-authored-by: Sora Morimoto <sora@morimoto.io> Co-authored-by: gengjiawen <technicalcute@gmail.com> Co-authored-by: Gustavo de León <alfonso.gus.deleon@gmail.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: Livia Rett <30511433+liviarett@users.noreply.github.com> Co-authored-by: TooTallNate <nathan@tootallnate.net> Co-authored-by: milani <mrtz.milani@googlemail.com> Co-authored-by: Psychless <jbeal24@gmail.com> Co-authored-by: mixu <mikito.takada@gmail.com> Co-authored-by: shtylman <shtylman@gmail.com> Co-authored-by: lloyd <lloyd@hilaiel.com> Co-authored-by: bolgovr <bolgovr@gmail.com> Co-authored-by: justinlatimer <justinlatimer@gmail.com> Co-authored-by: kkaefer <github@kkaefer.com> Co-authored-by: c4milo <camilo.aguilar@gmail.com> Co-authored-by: lperrin <laurent.perrin@gmail.com> Co-authored-by: oransel <oransel@yahoo.com> Co-authored-by: voodootikigod <chris@iterativedesigns.com> Co-authored-by: ehansin <ericc72@hotmail.com> Co-authored-by: xverges <xverges@gmail.com> Co-authored-by: Niggler <nirk.niggler@gmail.com> Co-authored-by: felquis <frgformenton@gmail.com> Co-authored-by: Dane Springmeyer <dane@dbsgeo.com> Co-authored-by: Alex Treppass <alextreppass@gmail.com> Co-authored-by: Luis Reis <luis.m.reis@gmail.com> Co-authored-by: lilo003 <lilo@c37.co> Co-authored-by: Syrian watermelon <AhmadBenmrad@users.noreply.github.com> Co-authored-by: Evan Su <hexacyanide@gmail.com> Co-authored-by: fov42550564 <42550564@qq.com> Co-authored-by: tcbeutler <tcbeutler@gmail.com> Co-authored-by: raztus <raztus@gmail.com> Co-authored-by: vweevers <mail@vincentweevers.nl> Co-authored-by: Andreas Brekken <a@abrkn.com> Co-authored-by: ralphtheninja <ralphtheninja@riseup.net> Co-authored-by: Zeke Sonxx <zeke@zekesonxx.com> Co-authored-by: Дмитрий Цветцих <dmitrycvet@gmail.com> Co-authored-by: Richard Winters <rik@mmogp.com> Co-authored-by: Mark Jeghers <jeghers@users.noreply.github.com> Co-authored-by: Marcin Cieślak <saper@saper.info> Co-authored-by: Dieter De Paepe <dieter.depaepe@gmail.com> Co-authored-by: Operations Research Engineering Software+ <alex@oresoftware.com> Co-authored-by: Flandre Scarlet <i@2333.moe> Co-authored-by: peter--bolier--zero <bolier@xs4all.nl> Co-authored-by: Nick Desaulniers <nickdesaulniers@users.noreply.github.com> Co-authored-by: Nicola Del Gobbo <nicoladelgobbo@gmail.com> Co-authored-by: Abdul Hameed <raza2022@gmail.com> Co-authored-by: xdf <xudafeng@126.com> Co-authored-by: Matt Hirsch <mhirsch@media.mit.edu> Co-authored-by: João Reis <reis@janeasystems.com> Co-authored-by: Bert Verhelst <verhelstbert@gmail.com> Co-authored-by: Mayank <9084735+mayank99@users.noreply.github.com> Co-authored-by: nineninesevenfour <75562299+nineninesevenfour@users.noreply.github.com> Co-authored-by: Cheng Zhao <zcbenz@gmail.com> Co-authored-by: Gar <gar+gh@danger.computer> Co-authored-by: csett86 <csett86@web.de> Co-authored-by: HeatonZ <905342035@qq.com> Co-authored-by: Rich Trott <rtrott@gmail.com> Co-authored-by: owl from hogvarts <47751812+owl-from-hogvarts@users.noreply.github.com> Co-authored-by: nlf <quitlahok@gmail.com> Co-authored-by: Mohamed-Elzohary <48733136+Mohamed-Elzohary@users.noreply.github.com> Co-authored-by: alexcfyung <alexcfyung@hotmail.com> Co-authored-by: Doni Rubiagatra <doni@zero-one-group.com> Co-authored-by: Nick Wang <nickwang14@gmail.com> Co-authored-by: Michael Dawson <mdawson@devrus.com> Co-authored-by: hubbergit <y.ahi@tms-bonn.de> Co-authored-by: Gaby Baghdadi <baghdadi@ca.ibm.com> Co-authored-by: Mr. Doge <42662615+FuPeiJiang@users.noreply.github.com> Co-authored-by: Kevin Adler <kadler@us.ibm.com> Co-authored-by: TrellixVulnTeam <kasimir.schulz@trellix.com> Co-authored-by: Luke Karrys <luke@lukekarrys.com> Co-authored-by: Raymond Zhao <7199958+rzhao271@users.noreply.github.com> Co-authored-by: Maksim Beliaev <beliaev.m.s@gmail.com> Co-authored-by: ravindraP20 <72969399+ravindraP20@users.noreply.github.com> Co-authored-by: Dennis Ameling <dennis@dennisameling.com> Co-authored-by: David Sanders <dsanders11@ucsbalum.com> Co-authored-by: Stefan Stojanovic <StefanStojanovic@users.noreply.github.com> Co-authored-by: James Cook <james@trainerroad.com> Co-authored-by: Rareș <6453351+raress96@users.noreply.github.com> Co-authored-by: Tim Perry <1526883+pimterry@users.noreply.github.com> Co-authored-by: Iulian Onofrei <5748627+revolter@users.noreply.github.com> Co-authored-by: Alex <aleksandrosansan@gmail.com> Co-authored-by: Richard Lau <rlau@redhat.com> Co-authored-by: Denis Bogomolov <denis@manwithbear.com>

DeeDeeG added 5 commits February 15, 2021 16:11

lib: drop Python 2 support in find-python.js

a9e3219

Python 2 isn't supported in gyp-next anymore, so we should stop finding/trying to use Python 2.

test: test find-python against supported Python 3

83bf041

Note: This commit is required for the previous commit to pass tests. I have left it separate for clarity during review of this PR. But it should be squashed with the previous commit if/when merging.

lib: use "-3" flag with py.exe launcher

9c0d980

Ensures the py.exe launcher only finds Python 3. (Only applies to Windows users.)

test: reject Python 2 in test-find-python

3c48f70

DeeDeeG force-pushed the find-python-only-find-python-3 branch from c889b78 to 3c48f70 Compare February 15, 2021 22:23

This comment has been minimized.

Sign in to view

cclauss reviewed Feb 16, 2021

View reviewed changes

lib/find-python.js Outdated Show resolved Hide resolved

cclauss reviewed Feb 16, 2021

View reviewed changes

lib/find-python.js Outdated Show resolved Hide resolved

lib: accept python '>=3.6.0' in find-python.js

f13747d

Co-authored-by: Christian Clauss <cclauss@me.com>

cclauss approved these changes Feb 16, 2021

View reviewed changes

DeeDeeG added 3 commits February 17, 2021 15:22

lib: use real Python 3 default install paths

d1f2aec

test: update Python path for "good guess" test

2562153

Note: This is required for the previous commit to pass tests. Please squash this with the previous commit if/when merging.

lib: use correct paths for 32-bit system-wide Python

d20f929

Paths are installed with a "-32" suffix: "Python[version]-32" I forgot to add the "-32" in previous commits. Note: It would be good to squash this with the previous two commits.

cclauss reviewed Feb 18, 2021

View reviewed changes

lib/find-python.js Outdated Show resolved Hide resolved

DeeDeeG added 2 commits February 18, 2021 16:53

lib: add system-wide Python path for 32-bit Windows

54041d8

32-bit Windows doesn't have a "Program Files (x86)" folder, only "Program Files".

lib: construct winDefaultLocations programatically

d0bcb5c

Reduces boilerplate code. Also, refactored to drop the dependency on the 'path' module.

mfonville reviewed Feb 19, 2021

View reviewed changes

cclauss requested review from joaocgreis, AshCripps, cjihrig, gengjiawen and targos February 20, 2021 10:07

gengjiawen approved these changes Feb 20, 2021

View reviewed changes

gengjiawen added the semver-major label Feb 20, 2021

gengjiawen requested a review from rvagg February 20, 2021 14:45

DeeDeeG commented Feb 21, 2021

View reviewed changes

test/test-find-python.js Outdated Show resolved Hide resolved

DeeDeeG commented Feb 21, 2021

View reviewed changes

DeeDeeG added 2 commits February 22, 2021 14:11

lib: fall back to os.userInfo() to get username

432447d

This should be a reliable fallback in case the environment variables have been deleted or unset.

DeeDeeG added 2 commits February 22, 2021 14:17

test: delete redundant Python 2 check

1711046

test: adjust expected number of tests

ec64f91

Note: This fixes an oversight in the previous commit, and is required for the previous commit to pass tests. Please squash this with the previous commit if/when merging.

mfonville approved these changes Feb 22, 2021

View reviewed changes

DeeDeeG mentioned this pull request Mar 20, 2021

Release proposal: v8.0.0 #2346

Merged

gengjiawen merged commit 1bd18f3 into nodejs:master Mar 26, 2021

This was referenced Apr 5, 2021

deps: node-gyp@8.0.0 npm/cli#3030

Closed

deps: node-gyp@8.0.0 npm/run-script#26

Closed

beltran-rubo mentioned this pull request Apr 12, 2021

Issue with non 'username' defined #2373

Closed

DeeDeeG mentioned this pull request May 6, 2021

lib: fail gracefully if we can't find the username #2375

Merged

4 tasks

DeeDeeG mentioned this pull request May 31, 2021

Wiki instructions for updating npm's bundled copy of node-gyp need updating #2416

Closed

DeeDeeG mentioned this pull request Aug 21, 2021

Debian Stretch-based images (including the default major version tags) ship with an older Python 3, which is incompatible with the latest node-gyp@8 nodejs/docker-node#1550

Closed

This was referenced Nov 1, 2021

[BUG] Can not build binary packages on Debian Stretch npm/run-script#40

Closed

Version 8 Lowest supported Python 3 should be 3.5 (instead of 3.6) #2540

Closed

carrodher mentioned this pull request Sep 1, 2022

[BUG]node:16-alpine3.16 and npm -v and other npm options return no output when using image in kubernetes npm/cli#5428

Open

2 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

lib: drop Python 2 support in find-python.js #2333

lib: drop Python 2 support in find-python.js #2333

DeeDeeG commented Feb 15, 2021 •

edited

Loading

This comment has been minimized.

cclauss left a comment

mfonville Feb 19, 2021

DeeDeeG Feb 20, 2021 •

edited

Loading

mfonville Feb 20, 2021

DeeDeeG Feb 20, 2021 •

edited

Loading

mfonville Feb 21, 2021

cclauss Feb 21, 2021

cclauss Feb 21, 2021

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 21, 2021 •

edited

Loading

cclauss Feb 21, 2021

cclauss Feb 21, 2021 •

edited

Loading

cjihrig Feb 21, 2021

DeeDeeG Feb 22, 2021

DeeDeeG Feb 22, 2021 •

edited

Loading

beltran-rubo Apr 9, 2021

DeeDeeG commented Feb 22, 2021 •

edited

Loading

cclauss commented Feb 22, 2021

mfonville commented Feb 22, 2021

cclauss commented Feb 22, 2021 •

edited

Loading

DeeDeeG commented Mar 20, 2021

rvagg commented Mar 22, 2021

gengjiawen commented Mar 26, 2021

cclauss commented Mar 26, 2021

lib: drop Python 2 support in find-python.js #2333

lib: drop Python 2 support in find-python.js #2333

Conversation

DeeDeeG commented Feb 15, 2021 • edited Loading

Checklist

Description of change

This comment has been minimized.

cclauss left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DeeDeeG Feb 20, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DeeDeeG Feb 20, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DeeDeeG Feb 22, 2021 • edited Loading

Choose a reason for hiding this comment

DeeDeeG Feb 22, 2021 • edited Loading

Choose a reason for hiding this comment

DeeDeeG Feb 22, 2021 • edited Loading

Choose a reason for hiding this comment

DeeDeeG Feb 21, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

cclauss Feb 21, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DeeDeeG Feb 22, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

DeeDeeG commented Feb 22, 2021 • edited Loading

cclauss commented Feb 22, 2021

mfonville commented Feb 22, 2021

cclauss commented Feb 22, 2021 • edited Loading

DeeDeeG commented Mar 20, 2021

rvagg commented Mar 22, 2021

gengjiawen commented Mar 26, 2021

cclauss commented Mar 26, 2021

DeeDeeG commented Feb 15, 2021 •

edited

Loading

DeeDeeG Feb 20, 2021 •

edited

Loading

DeeDeeG Feb 20, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG Feb 21, 2021 •

edited

Loading

cclauss Feb 21, 2021 •

edited

Loading

DeeDeeG Feb 22, 2021 •

edited

Loading

DeeDeeG commented Feb 22, 2021 •

edited

Loading

cclauss commented Feb 22, 2021 •

edited

Loading