Skip to content

Latest commit

 

History

History
644 lines (510 loc) · 28 KB

COLLABORATOR_GUIDE.md

File metadata and controls

644 lines (510 loc) · 28 KB
Raw

Node.js Collaborator Guide

Contents

This document contains information for Collaborators of the Node.js project regarding maintaining the code, documentation and issues.

Collaborators should be familiar with the guidelines for new contributors in CONTRIBUTING.md and also understand the project governance model as outlined in GOVERNANCE.md.

Issues and Pull Requests

Courtesy should always be shown to individuals submitting issues and pull requests to the Node.js project.

Collaborators should feel free to take full responsibility for managing issues and pull requests they feel qualified to handle, as long as this is done while being mindful of these guidelines, the opinions of other Collaborators and guidance of the CTC.

Collaborators may close any issue or pull request they believe is not relevant for the future of the Node.js project. Where this is unclear, the issue should be left open for several days to allow for additional discussion. Where this does not yield input from Node.js Collaborators or additional evidence that the issue has relevance, the issue may be closed. Remember that issues can always be re-opened if necessary.

See "Who to CC in issues"

Accepting Modifications

All modifications to the Node.js code and documentation should be performed via GitHub pull requests, including modifications by Collaborators and CTC members.

All pull requests must be reviewed and accepted by a Collaborator with sufficient expertise who is able to take full responsibility for the change. In the case of pull requests proposed by an existing Collaborator, an additional Collaborator is required for sign-off.

In some cases, it may be necessary to summon a qualified Collaborator to a pull request for review by @-mention.

If you are unsure about the modification and are not prepared to take full responsibility for the change, defer to another Collaborator.

Before landing pull requests, sufficient time should be left for input from other Collaborators. Leave at least 48 hours during the week and 72 hours over weekends to account for international time differences and work schedules. Trivial changes (e.g. those which fix minor bugs or improve performance without affecting API or causing other wide-reaching impact) may be landed after a shorter delay.

For non-breaking changes, if there is no disagreement amongst Collaborators, a pull request may be landed given appropriate review. Where there is discussion amongst Collaborators, consensus should be sought if possible. The lack of consensus may indicate the need to elevate discussion to the CTC for resolution (see below).

Breaking changes (that is, pull requests that require an increase in the major version number, known as semver-major changes) must be elevated for review by the CTC. This does not necessarily mean that the PR must be put onto the CTC meeting agenda. If multiple CTC members approve (LGTM) the PR and no Collaborators oppose the PR, it can be landed. Where there is disagreement among CTC members or objections from one or more Collaborators, semver-major pull requests should be put on the CTC meeting agenda.

All bugfixes require a test case which demonstrates the defect. The test should fail before the change, and pass after the change.

All pull requests that modify executable code should be subjected to continuous integration tests on the project CI server.

Useful CI Jobs

  • node-test-pull-request is the standard CI run we do to check Pull Requests. It triggers node-test-commit, which runs the build-ci and test-ci targets on all supported platforms.

  • node-test-linter only runs the linter targets, which is useful for changes that only affect comments or documentation.

  • citgm-smoker uses CitGM to allow you to run npm install && npm test on a large selection of common modules. This is useful to check whether a change will cause breakage in the ecosystem. To test Node.JS ABI changes you can run citgm-abi-smoker.

  • node-stress-single-test is designed to allow one to run a group of tests over and over on a specific platform to confirm that the test is reliable.

  • node-test-commit-v8-linux is designed to allow validation of changes to the copy of V8 in the Node.js tree by running the standard V8 tests. It should be run whenever the level of V8 within Node.js is updated or new patches are floated on V8.

Internal vs. Public API

Due to the nature of the JavaScript language, it can often be difficult to establish a clear distinction between which parts of the Node.js implementation represent the "public" API Node.js users should assume to be stable and which are considered part of the "internal" implementation detail of Node.js itself. A general rule of thumb has been to base the determination off what functionality is actually documented in the official Node.js API documentation. However, it has been repeatedly demonstrated that either the documentation does not completely cover implemented behavior or that Node.js users have come to rely heavily on undocumented aspects of the Node.js implementation.

While there are numerous exceptions, the following general rules should be followed to determine which aspects of the Node.js API are considered "internal":

  • Any and all functionality exposed via process.binding(...) is considered to be internal and not part of the Node.js Public API.
  • Any and all functionality implemented in lib/internal/**/*.js that is not re-exported by code in lib/*.js, or is not documented as part of the Node.js Public API, is considered to be internal.
  • Any object property or method whose key is a non-exported Symbol is considered to be an internal property.
  • Any object property or method whose key begins with the underscore _ prefix, and is not documented as part of the Node.js Public API, is considered to be an internal property.
  • Any object, property, method, argument, behavior, or event not documented in the Node.js documentation is considered to be internal.
  • Any native C/C++ APIs/ABIs exported by the Node.js *.h header files that are hidden behind the NODE_WANT_INTERNALS flag are considered to be internal.

Exception to each of these points can be made if use or behavior of a given internal API can be demonstrated to be sufficiently relied upon by the Node.js ecosystem such that any changes would cause too much breakage. The threshold for what qualifies as "too much breakage" is to be decided on a case-by-case basis by the CTC.

If it is determined that a currently undocumented object, property, method, argument, or event should be documented, then a pull request adding the documentation is required in order for it to be considered part of the "public" API.

Making a determination about whether something should be documented can be difficult and will need to be handled on a case-by-case basis. For instance, if one documented API cannot be used successfully without the use of a second currently undocumented API, then the second API should be documented. If using an API in a manner currently undocumented achieves a particular useful result, a decision will need to be made whether or not that falls within the supported scope of that API; and if it does, it should be documented.

Breaking changes to internal elements are permitted in semver-patch or semver-minor commits but Collaborators should take significant care when making and reviewing such changes. Before landing such commits, an effort must be made to determine the potential impact of the change in the ecosystem by analyzing current use and by validating such changes through ecosystem testing using the Canary in the Goldmine tool. If a change cannot be made without ecosystem breakage, then CTC review is required before landing the change as anything less than semver-major.

If a determination is made that a particular internal API (for instance, an underscore _ prefixed property) is sufficiently relied upon by the ecosystem such that any changes may break user code, then serious consideration should be given to providing an alternative Public API for that functionality before any breaking changes are made.

Breaking Changes

Backwards-incompatible changes may land on the master branch at any time after sufficient review by collaborators and approval of at least two CTC members.

Examples of breaking changes include, but are not necessarily limited to, removal or redefinition of existing API arguments, changing return values (except when return values do not currently exist), removing or modifying existing properties on an options argument, adding or removing errors, changing error messages in any way, altering expected timing of an event (e.g. moving from sync to async responses or vice versa), and changing the non-internal side effects of using a particular API.

With a few notable exceptions outlined below, when backwards incompatible changes to a Public API are necessary, the existing API must be deprecated first and the new API either introduced in parallel or added after the next major Node.js version following the deprecation as a replacement for the deprecated API. In other words, as a general rule, existing Public APIs must not change (in a backwards incompatible way) without a deprecation.

Exception to this rule is given in the following cases:

  • Adding or removing errors thrown or reported by a Public API;
  • Changing error messages;
  • Altering the timing and non-internal side effects of the Public API.

Such changes must be handled as semver-major changes but MAY be landed without a Deprecation cycle.

From time-to-time, in particularly exceptional cases, the CTC may be asked to consider and approve additional exceptions to this rule.

Purely additive changes (e.g. adding new events to EventEmitter implementations, adding new arguments to a method in a way that allows existing code to continue working without modification, or adding new properties to an options argument) are handled as semver-minor changes.

Note that errors thrown, along with behaviors and APIs implemented by dependencies of Node.js (e.g. those originating from V8) are generally not under the control of Node.js and therefore are not directly subject to this policy. However, care should still be taken when landing updates to dependencies when it is known or expected that breaking changes to error handling may have been made. Additional CI testing may be required.

When breaking changes actually break things

Breaking changes are difficult primarily because they change the fundamental assumptions a user of Node.js has when writing their code and can cause existing code to stop functioning as expected -- costing developers and users time and energy to fix.

Because breaking (semver-major) changes are permitted to land in master at any time, it should be understood and expected that at least some subset of the user ecosystem may be adversely affected in the short term when attempting to build and use Node.js directly from master. This potential instability is precisely why Node.js offers distinct Current and LTS release streams that offer explicit stability guarantees.

Specifically:

  • Breaking changes should never land in Current or LTS except when:
    • Resolving critical security issues.
    • Fixing a critical bug (e.g. fixing a memory leak) requires a breaking change.
    • There is CTC consensus that the change is required.
  • If a breaking commit does accidentally land in a Current or LTS branch, an attempt to fix the issue will be made before the next release; If no fix is provided then the commit will be reverted.

When any change is landed in master, and it is determined that the such changes do break existing code, a decision may be made to revert those changes either temporarily or permanently. However, the decision to revert or not can often be based on many complex factors that are not easily codified. It is also possible that the breaking commit can be labeled retroactively as a semver-major change that will not be backported to Current or LTS branches.

Reverting commits

Commits are reverted with git revert <HASH>, or git revert <FROM>..<TO> for multiple commits. Commit metadata and the reason for the revert should be appended. Commit message rules about line length and subsystem can be ignored. A Pull Request should be raised and approved like any other change.

Deprecations

Deprecation refers to the identification of Public APIs that should no longer be used and that may be removed or modified in non-backwards compatible ways in a future major release of Node.js. Deprecation may be used with internal APIs if there is expected impact on the user community.

Node.js uses three fundamental Deprecation levels:

  • Documentation-Only Deprecation refers to elements of the Public API that are being staged for deprecation in a future Node.js major release. An explicit notice indicating the deprecated status is added to the API documentation but no functional changes are implemented in the code. There will be no runtime deprecation warning emitted for such deprecations.

  • Runtime Deprecation refers to the use of process warnings emitted at runtime the first time that a deprecated API is used. A command-line switch can be used to escalate such warnings into runtime errors that will cause the Node.js process to exit. As with Documentation-Only Deprecation, the documentation for the API must be updated to clearly indicate the deprecated status.

  • End-of-life refers to APIs that have gone through Runtime Deprecation and are ready to be removed from Node.js entirely.

Documentation-Only Deprecations may be handled as semver-minor or semver-major changes. Such deprecations have no impact on the successful operation of running code and therefore should not be viewed as breaking changes.

Runtime Deprecations and End-of-life APIs (internal or public) must be handled as semver-major changes unless there is CTC consensus to land the deprecation as a semver-minor.

All Documentation-Only and Runtime deprecations will be assigned a unique identifier that can be used to persistently refer to the deprecation in documentation, emitted process warnings, or errors thrown. Documentation for these identifiers will be included in the Node.js API documentation and will be immutable once assigned. Even if End-of-Life code is removed from Node.js, the documentation for the assigned deprecation identifier must remain in the Node.js API documentation.

A "Deprecation cycle" is one full Node.js major release during which an API has been in one of the three Deprecation levels. (Note that Documentation-Only Deprecations may land in a Node.js minor release but must not be upgraded to a Runtime Deprecation until the next major release.)

No API can be moved to End-of-life without first having gone through a Runtime Deprecation cycle.

A best effort will be made to communicate pending deprecations and associated mitigations with the ecosystem as soon as possible (preferably before the pull request adding the deprecation lands in master). All deprecations included in a Node.js release should be listed prominently in the "Notable Changes" section of the release notes.

Involving the CTC

Collaborators may opt to elevate pull requests or issues to the CTC for discussion by assigning the ctc-review label. This should be done where a pull request:

  • has a significant impact on the codebase,
  • is inherently controversial; or
  • has failed to reach consensus amongst the Collaborators who are actively participating in the discussion.

The CTC should serve as the final arbiter where required.

Landing Pull Requests

  • Please never use GitHub's green "Merge Pull Request" button.
    • If you do, please force-push removing the merge.
    • Reasons for not using the web interface button:
      • The merge method will add an unnecessary merge commit.
      • The rebase & merge method adds metadata to the commit title.
      • The rebase method changes the author.
      • The squash & merge method has been known to add metadata to the commit title.
      • If more than one author has contributed to the PR, only the latest author will be considered during the squashing.

Always modify the original commit message to include additional meta information regarding the change process:

  • A PR-URL: line that references the full GitHub URL of the original pull request being merged so it's easy to trace a commit back to the conversation that led up to that change.
  • A Fixes: X line, where X either includes the full GitHub URL for an issue, and/or the hash and commit message if the commit fixes a bug in a previous commit. Multiple Fixes: lines may be added if appropriate.
  • A Refs: line referencing a URL for any relevant background.
  • A Reviewed-By: Name <email> line for yourself and any other Collaborators who have reviewed the change.
    • Useful for @mentions / contact list if something goes wrong in the PR.
    • Protects against the assumption that GitHub will be around forever.

Review the commit message to ensure that it adheres to the guidelines outlined in the contributing guide.

See the commit log for examples such as this one if unsure exactly how to format your commit messages.

Additionally:

  • Double check PRs to make sure the person's full name and email address are correct before merging.
  • Except when updating dependencies, all commits should be self contained (meaning every commit should pass all tests). This makes it much easier when bisecting to find a breaking change.

Technical HOWTO

Clear any am/rebase that may already be underway.

$ git am --abort
$ git rebase --abort

Checkout proper target branch

$ git checkout master

Update the tree (assumes your repo is set up as detailed in CONTRIBUTING.md)

$ git fetch upstream
$ git merge --ff-only upstream/master

Apply external patches

$ curl -L https://github.com/nodejs/node/pull/xxx.patch | git am --whitespace=fix

If the merge fails even though recent CI runs were successful, then a 3-way merge may be required. In this case try:

$ git am --abort
$ curl -L https://github.com/nodejs/node/pull/xxx.patch | git am -3 --whitespace=fix

If the 3-way merge succeeds you can proceed, but make sure to check the changes against the original PR carefully and build/test on at least one platform before landing. If the 3-way merge fails, then it is most likely that a conflicting PR has landed since the CI run and you will have to ask the author to rebase.

Check and re-review the changes

$ git diff upstream/master

Check number of commits and commit messages

$ git log upstream/master...master

If there are multiple commits that relate to the same feature or one with a feature and separate with a test for that feature, you'll need to use squash or fixup:

$ git rebase -i upstream/master

This will open a screen like this (in the default shell editor):

pick 6928fc1 crypto: add feature A
pick 8120c4c add test for feature A
pick 51759dc feature B
pick 7d6f433 test for feature B

# Rebase f9456a2..7d6f433 onto f9456a2
#
# Commands:
#  p, pick = use commit
#  r, reword = use commit, but edit the commit message
#  e, edit = use commit, but stop for amending
#  s, squash = use commit, but meld into previous commit
#  f, fixup = like "squash", but discard this commit's log message
#  x, exec = run command (the rest of the line) using shell
#
# These lines can be re-ordered; they are executed from top to bottom.
#
# If you remove a line here THAT COMMIT WILL BE LOST.
#
# However, if you remove everything, the rebase will be aborted.
#
# Note that empty commits are commented out

Replace a couple of picks with fixup to squash them into a previous commit:

pick 6928fc1 crypto: add feature A
fixup 8120c4c add test for feature A
pick 51759dc feature B
fixup 7d6f433 test for feature B

Replace pick with reword to change the commit message:

reword 6928fc1 crypto: add feature A
fixup 8120c4c add test for feature A
reword 51759dc feature B
fixup 7d6f433 test for feature B

Save the file and close the editor. You'll be asked to enter a new commit message for that commit. This is a good moment to fix incorrect commit logs, ensure that they are properly formatted, and add Reviewed-By lines.

Run tests (make -j4 test or vcbuild test). Even though there was a successful continuous integration run, other changes may have landed on master since then, so running the tests one last time locally is a good practice.

Time to push it:

$ git push upstream master

  • Optional: Force push the amended commit to the branch you used to open the pull request. If your branch is called bugfix, then the command would be git push --force-with-lease origin master:bugfix. When the pull request is closed, this will cause the pull request to show the purple merged status rather than the red closed status that is usually used for pull requests that weren't merged. Only do this when landing your own contributions.

  • Close the pull request with a "Landed in <commit hash>" comment. If your pull request shows the purple merged status then you should still add the "Landed in .." comment if you added multiple commits.

Troubleshooting

Sometimes, when running git push upstream master, you may get an error message like this:

To https://github.com/nodejs/node
 ! [rejected]              master -> master (fetch first)
error: failed to push some refs to 'https://github.com/nodejs/node'
hint: Updates were rejected because the remote contains work that you do
hint: not have locally. This is usually caused by another repository pushing
hint: to the same ref. You may want to first integrate the remote changes
hint: (e.g., 'git pull ...') before pushing again.
hint: See the 'Note about fast-forwards' in 'git push --help' for details.

That means a commit has landed since your last rebase against upstream/master. To fix this, fetch, rebase, run the tests again (to make sure no interactions between your changes and the new changes cause any problems), and push again:

git fetch upstream
git rebase upstream/master
make -j4 test
git push upstream master

I Just Made a Mistake

  • Ping a CTC member.
  • #node-dev on freenode
  • With git, there's a way to override remote trees by force pushing (git push -f). This should generally be seen as forbidden (since you're rewriting history on a repository other people are working against) but is allowed for simpler slip-ups such as typos in commit messages. However, you are only allowed to force push to any Node.js branch within 10 minutes from your original push. If someone else pushes to the branch or the 10 minute period passes, consider the commit final.
    • Use --force-with-lease to minimize the chance of overwriting someone else's change.
    • Post to #node-dev (IRC) if you force push.

Long Term Support

What is LTS?

Long Term Support (often referred to as LTS) guarantees application developers a 30 month support cycle with specific versions of Node.js.

You can find more information in the full LTS plan.

How does LTS work?

Once a Current branch enters LTS, changes in that branch are limited to bug fixes, security updates, possible npm updates, documentation updates, and certain performance improvements that can be demonstrated to not break existing applications. Semver-minor changes are only permitted if required for bug fixes and then only on a case-by-case basis with LTS WG and possibly Core Technical Committee (CTC) review. Semver-major changes are permitted only if required for security related fixes.

Once a Current branch moves into Maintenance mode, only critical bugs, critical security fixes, and documentation updates will be permitted.

Landing semver-minor commits in LTS

The default policy is to not land semver-minor or higher commits in any LTS branch. However, the LTS WG or CTC can evaluate any individual semver-minor commit and decide whether a special exception ought to be made. It is expected that such exceptions would be evaluated, in part, on the scope and impact of the changes on the code, the risk to ecosystem stability incurred by accepting the change, and the expected benefit that landing the commit will have for the ecosystem.

Any collaborator who feels a semver-minor commit should be landed in an LTS branch should attach the lts-agenda label to the pull request. The LTS WG will discuss the issue and, if necessary, will escalate the issue up to the CTC for further discussion.

How are LTS Branches Managed?

There are currently two LTS branches: v6.x and v4.x. Each of these is paired with a "staging" branch: v6.x-staging and v4.x-staging.

As commits land in master, they are cherry-picked back to each staging branch as appropriate. If the commit applies only to the LTS branch, the PR must be opened against the staging branch. Commits are selectively pulled from the staging branch into the LTS branch only when a release is being prepared and may be pulled into the LTS branch in a different order than they were landed in staging.

Any collaborator may land commits into a staging branch, but only the release team should land commits into the LTS branch while preparing a new LTS release.

How can I help?

When you send your pull request, consider including information about whether your change is breaking. If you think your patch can be backported, please feel free to include that information in the PR thread. For more information on backporting, please see the backporting guide.

Several LTS related issue and PR labels have been provided:

  • lts-watch-v6.x - tells the LTS WG that the issue/PR needs to be considered for landing in the v6.x-staging branch.
  • lts-watch-v4.x - tells the LTS WG that the issue/PR needs to be considered for landing in the v4.x-staging branch.
  • land-on-v6.x - tells the release team that the commit should be landed in a future v6.x release
  • land-on-v4.x - tells the release team that the commit should be landed in a future v4.x release

Any collaborator can attach these labels to any PR/issue. As commits are landed into the staging branches, the lts-watch- label will be removed. Likewise, as commits are landed in a LTS release, the land-on- label will be removed.

Collaborators are encouraged to help the LTS WG by attaching the appropriate lts-watch- label to any PR that may impact an LTS release.

How is an LTS release cut?

When the LTS working group determines that a new LTS release is required, selected commits will be picked from the staging branch to be included in the release. This process of making a release will be a collaboration between the LTS working group and the Release team.