If you have concerns about your feature not fitting into this process, or any other questions, general concerns or discussion on this page or the process it describes, please e-mail email@example.com.
You do not need to use this process if your change does not affect web API behavior to the point that developers need to be aware of it. (e.g. no significant behavioral changes, and no web-facing API changes.) The rest of this document doesn’t apply to this type of change, although such features might still have to go through a different launch process. Large projects should have public design docs that are also shared on firstname.lastname@example.org (or chromium-dev@, for projects that have significant parts outside of Blink) for feedback (this is also a good way to get the attention of other relevant leads).
Frequently asked questions
Q: Do I need any of this if my project is just refactoring or re-architecting the code? Do the API owners need to be involved?
A: No. The API owners oversee the process of shipping web-exposed API changes. They are not necessarily leads or overseers of any of the code. Instead, you should get the buy-in of the leads of the code areas touched by your project. If there may be side effects of your change, you should follow the "Architectural change" process below. In addition, such large projects should have public design docs that are also shared on email@example.com (or firstname.lastname@example.org, for projects that have significant parts outside of third_party/blink) for feedback (this is also a good way to get the attention of relevant leads you might not have thought of).
For code-related questions, you can email email@example.com in addition to blink-dev@ as a catch-all option when the code ownership is not clear or the feature needs large-scale refactoring changes.
Q: What if I want to add some code to third_party/blink that is for a non-web-exposed feature? Is that allowed?
A: In general, no. On a case-by-case basis an exception could be given if there is no other reasonable way to implement the feature. Ask for permission from leads of the code area as well as the API owners. (The API owners need to be involved only to help understand if the feature really is not web-exposed; this can be a very subtle question.)
Q: I am not sure of the right approach for my feature. What should I do?
A: Please reach out to the API owners for help! While they are not gatekeepers for everything, they are very happy to give advice and unblock your feature. An email to firstname.lastname@example.org is the best way; if a public-facing email is not possible, please email the API owners directly.
The Feature Types
The first thing you will need to do is identify what type of feature you are building:
New feature incubation
This is the normal path when we are incubating and defining new features from scratch - e.g., most Fugu features follow this pattern, and any feature where we start without an already-defined standard for the feature. This also covers changes that are extending existing features (e.g., defining a new value and behavior for an existing standard feature). This type of feature has the most associated process steps, as it is charting new territory.
Implementation of existing standard
This type of feature has a lighter-weight “fast track” process, but this process can only be used when the feature has already been defined in a consensus standards document - e.g. a W3C Working Group has already agreed on the design, it has already been merged into a WHATWG standard, or the feature has already been implemented in another engine.
Removal of already-shipped features.
Web developer facing change to existing code
This is generally a public service announcement- “This is a web-developer-facing change to existing code without API changes, but you may see side effects.” This may be due to large-scale code refactoring or rewriting, where the goal is to cause no behavioral changes (but due to scope of change, side effects are likely), or this may encompass changes to current code that fix a bug or implement new changes to the spec without changes to the API shape itself.
It is possible to change types later in the process - for example, if you start out implementing an already existing standard, but discover you need to incubate a new API during the process, you can change the feature type and move back stages. Note that there are few strictly required gates to the Chromium process (e.g., 3 LGTMs from API owners on an intent-to-ship) - particularly in the earlier stages we want to encourage experimentation. However, there are required fields for most stages in the ChromeStatus tool, and it’s important to consider all the steps listed below; this will maximize your chance of success on an intent-to-ship and reduce the risk of having to redo parts of your design or implementation.
The Chromium process to launch a new feature
Step 0: Create a ChromeStatus entry, and choose your feature type.
For all types of features, the first step is to create a ChromeStatus entry. Go to https://www.chromestatus.com/features, ensure you are logged in (see the upper right corner), and click “Add new feature”. If you do no have access to create a new feature, please send an email to email@example.com to request access. Follow the directions to name your feature and give a short summary, and select the appropriate feature type.
For Chrome, some launches will require a formal Chrome launch review (especially if your feature has security, privacy, legal, or UI implications). You can work with a Google counterpart to get all launch approvals. We are working on making this process more open and transparent outside Google (new tooling is in progress).
From this point on, the process changes a little depending on the type of feature you’re adding.
For New Feature Incubations
Step 1: Incubating: Write up use cases and scenarios in an explainer
Proceed to the "Start Incubating" stage in Chrome Status.
Write up the use cases and scenarios for the feature in an explainer, which is typically written in Markdown. Then host your explainer at a public URL, typically in a public personal Github repo, and put a link to your public explainer in the feature entry.
We have a program for to provide mentorship for specification writing; If you are a Googler you must file a request for a spec mentor, and ask them to review this early explainer before proceeding. If you are not a Googler, you are welcome to make use of this but not required to.
Then kick off "standards incubation" by proposing your explainer to a relevant incubation venue or working group and socializing the problem with other vendors and developers. For web features, the WICG is a good default, but ask your spec mentor if you need help picking a venue. Enter a reference to the public proposal in “Initial Public Proposal” field. It’s also a good idea to discuss your idea with the team/team lead (TL) or area expert in the feature area, prior to checking in code in the area.
After some discussion with the community, start sketching out a proposed solution in your (public) explainer, including sample code for using your API design. You may wish to review this with your specification mentor before proceeding to prototyping.
Step 2: Prototyping
Proceed to the “Start Prototyping” stage in ChromeStatus - this will generate an “Intent to Prototype” mail for you. Send that email to blink-dev and start checking in prototype code to Chromium under a runtime flag. You should do your detailed API design in the open, in your public repository, and response to feedback filed there. You should continue pushing for public engagement (from other vendors and web developers), and to move into an incubation venue or working group if you haven’t already. During this stage, you should expand on the explainer with a full design doc (this may also have implementation-specific details), and consider creating a specification (or writing up a pull request with changes to an existing spec).
Note that any CLs landing at this stage should be behind a feature flag. Consider adding a UseCounter to track usage of your new feature in the wild, and be sure to write integration tests for your feature as web-platform-tests as you go. Continue to work with your API mentor if there are any design changes.
Ensure you have an API overview and descriptions for all IDL methods and properties (these are probably in your specification or explainer, but developers will need them to try your feature out), and at least a basic sample.
As soon as you have a functional and reasonably complete implementation of your initial design ready for developers to try out behind a flag, proceed to the next step.
Step 3: Feature Complete behind a feature flag: iteration on design
Once you have a functional and reasonably complete feature implementation available as a runtime enabled feature, we recommend (but don't require) that you request an Early Design Review from the TAG and proceed to the “Dev Trials” stage in ChromeStatus (note also exceptions enumerated here). This will generate a “Ready for Trial” email that you should send to blink-dev to notify the community they can try out the feature. At this point, you should consider asking other browser vendors and the web developer community for signals on their opinion of the API.
This is the main iterating stage of feature development and helps you assess product-market-fit early on before you corner yourself (does your API address a problem with meaningful demand? did we get the ergonomics right?). You may wish to send more than one Ready for Trial emails, if you make substantial changes to the API shape while iterating. You should work with the TAG to complete their review and address any issues raised during this stage, and should address any issues raised by other horizontal reviews (accessibility, privacy, internationalization, etc.).
Step 4: Widen review
Once you believe you have addressed all major open issues, you should proceed to the “Evaluating readiness to ship” stage in ChromeStatus.
By this point, your explainer and other public specification work need to be hosted in an incubation venue (or a full working group), in order to assure contributors that IP concerns are being handled appropriately. Any W3C community group is acceptable, as is publication as an IETF internet draft. For web features, the WICG is a good default, but ask your spec mentor if you need help picking a venue.
If you haven't already received signals on their opinion of the API from other browser vendors and the web developer community, now is the time to pursue getting those signals.
You should also work with the documentation team to ensure your features will be documented when shipped, and estimate when (in what milestone) you would like to target shipping.
You should also decide if an Origin Trial would help gather significant data for your feature.
Now that you think your overall design is relatively settled (or just after starting your Origin Trial), if you haven't already, start writing your specification. Talk to your spec mentor about how long you should expect this to take. As a very rough guess, estimate at least a month of full-time work, but that can take more calendar time than the 12-16 weeks of an Origin Trial. Starting the spec-writing process too late might delay your feature launch.
Once you have a complete specification, request a Specification Review from the TAG (except in the cases noted here). You should submit this at least a month ahead of sending an Intent to Ship, to give the TAG sufficient time for meaningful feedback.
Step 5 (Optional): Origin Trial
If you want to gather data on the usability of your feature that an Origin Trial can help collect, proceed to the “Origin Trial” stage in ChromeStatus and fill out the required fields detailing what you hope to learn from the origin trial. This will generate an Intent to Experiment mail that you should send to blink-dev. After receiving at least one LGTM from the API owners, you can proceed with your origin trial release. Collect data and respond to any issues. From here, you may wish to return to Dev Trials, proceed to Prepare to Ship, or park the feature.
As noted above, a specification is recommended but not required for an Origin Trial.
Please note that Origin Trials are not exempt from requiring cross-functional approvals from the Chrome launch review process.
Depending on your feature and your experimentation goals, running an experiment via Finch on a percentage of the stable or beta populations may be useful (example). In these cases, an Intent to Experiment explaining why this non-typical path is requested and the corresponding LGTM(s) are still required before proceeding.
An initial origin trial for a feature may only run for 6 milestones of Chromium. Each request to extend beyond that limit may only be for 3 milestones at a time, and will not be approved unless substantial progress is demonstrated in all of these areas:
- Draft spec (early draft is ok, but must be spec-like and associated with the appropriate standardization venue, or WICG)
- TAG review (see exceptions)
- bit.ly/blink-signals requests
- Outreach for feedback from the spec community
- WPT tests
Each subsequent request to extend an origin trial must provide substantial additional progress on top of the previous extension request.
[Note: the required breaking period was removed in April 2022. This removal will run
for 12 months, after which the API owners will consider making the change permanent.]
If an Origin Trial happens, then there is a required breaking
before shipping in step 4. If you wish to skip the breaking period, meaning that
sites participating in the Origin Trial will not see an interruption in support
for the feature between Origin Trial and launch (*), you may request an
exception. The process to do so is to include this request in your Intent to
Ship email. In the request, you must show clear evidence that developers engaged
with the Origin Trial and that their concerns were taken into account in the
final API design and implementation. LGTMs for the Intent to Ship imply approval
of the request.
(*) "Not see an interruption" means that if the origin trial ends at milestone N, and the feature is shipped in milestone N+1, sites opting into the origin trial will continue to be able to use the feature on Chromium milestone N up to (and even after, for those users who have not upgraded) N+1 ships.
Step 6: Prepare to Ship
By this stage, you need to have a complete specification available that matches what you have implemented, and you should have given the TAG at least a month to comment on that specification. If you are a Googler you should get a final spec review from your spec mentor, and discuss options for moving your spec to a final standardization venue. You should get final signoff from Documentation.
If the TAG has commented on your review, and the state of the review is anything
the API Owners should be able to see evidence that you've seriously and
comprehensively engaged with their comments, and tried to resolve any concerns.
If your specification is still in an incubation venue and not a working group, propose that the feature migrate to a working group. There's no requirement that a working group adopt the feature before Chromium ships it, but if we're ready to ship it, we need to declare publicly that we think incubation is finished. Ask your spec mentor for help with this.
Update ChromeStatus with:
- a target milestone for shipping (and remember to keep this updated, if things change) and
- any changes in vendor signals.
Proceed to the “Prepare to Ship” stage in ChromeStatus; this will generate an Intent to Ship mail that you should send to blink-dev. This will spark a conversation with the API owners; address any feedback from them, and once you get 3 LGTMs from the API owners, you may enable the feature by default. You can learn more about the policies and guidelines the API owners evaluate here. Requirements for new API owners are here. Email firstname.lastname@example.org or reach out to one of the API owners if there is no open/unaddressed feedback and you are still blocked on LGTMs after 5 days.
Once you have shipped your feature, proceed to the "Ship" stage in ChromeStatus.
The approval status of various stages of the intent process is tracked by the API owners in this spreadsheet.
Implementations of already-defined consensus-based standards
Step 1: Write up use cases and scenarios, start coding
Fill out the “Motivation” section with a brief summary, and then write up the use cases and scenarios. If this is a large feature, or a combination of multiple attributes/properties/methods/events, you may wish to do this in a separate explainer file.
Proceed to the “Start Prototyping” stage in ChromeStatus - this will generate an “Intent to Prototype” mail for you. Send that email to blink-dev and start checking in prototype code to Chromium as runtime enabled features. Ensure there are adequate web-platform-tests for this feature. Also ensure you have an API overview and descriptions for all IDL methods and properties (this is probably already in the consensus standard specification, or even on MDN), and at least a basic sample.
As soon as you have a functional and reasonably complete implementation of the feature ready for developers to try out under a flag, proceed to the next step.
Step 2: Feature Complete behind a flag and implementation refinement
If the TAG has not already reviewed the consensus specification, request a Specification Review and proceed to the “Dev Trials” stage in ChromeStatus. This will generate a “Ready for Trial” email that you should send to blink-dev to notify the community they can try out the feature.
After you have addressed any issues that the community finds, you should proceed to the “Evaluating readiness to ship” stage in ChromeStatus. You should also work with the documentation team to ensure your feature will be documented when shipped, and estimate when (in what milestone) you would like to target shipping. You should also decide if an Origin Trial would help gather significant data for your feature.
Step 3 (Optional): Origin Trial
The Origin Trial guidelines for new feature incubations applies here as well.
Step 4: Prepare to Ship
You should update ChromeStatus with a target milestone for shipping (and remember to keep this updated, if things change). If you are working with DevRel on documentation, this is the time to finish that work and update for any changes in vendor signals.
Proceed to the “Prepare to Ship” stage in ChromeStatus; this will generate an Intent to Ship mail that you should send to blink-dev. This will spark a conversation with the API owners; address any feedback from them, and once you get 3 LGTMs from the API owners, you may enable the feature by default. See here for the principles the API owners will apply in evaluating whether the feature is mature enough to ship. (API owners requirements are listed here - email email@example.com or reach out to one of the API owners if no open/unaddressed feedback and you are still blocked on LGTMs after 5 days.)
Once you have shipped your feature, proceed to the "Ship" stage in ChromeStatus.
Lessons from the first years of deprecations and removals (thread)
We should weigh the benefits of removing an API more against the cost it has. Percent of page views by itself is not the only metric we care about.
The cost of removing an API is not accurately reflected by the UseCounter for older, widely implemented APIs. It's more likely that there's a longer-tail of legacy content that we're breaking.
We shouldn't remove APIs that have small value on the path towards a removal that has significant value. Getting rid of attribute nodes *is* valuable and would benefit the platform. Getting rid of half the attribute node methods is not. So we should evaluate the usage of all the APIs we need to remove together in order to get there. Also, if we remove them, we should remove them all in the same release. Breaking people once is better than breaking them repeatedly in small ways.
We should be more hesitant to remove older, widely implemented APIs.
For cases where we're particularly concerned about the compatibility hit, we should do the removal behind a flag so that we can easily re-enable the API on stable as we don't know the compat hit until the release has been on stable for a couple weeks.
Enterprise users have additional difficulties reacting to breaking changes, but we also have additional tools to help them. See shipping changes that are enterprise-friendly for best practices.
High-usage APIs may require much more work to land successfully. See here for a good example of how this worked in practice with the deprecation and removal of the Web Components v0 APIs.
Step 1: Write up motivation
Deprecations and removals must have strong reasons, explicitly balanced against the cost of removal. Deprecations should be clear and actionable for developers. First, read the guidelines for deprecating a feature. Measure feature usage in the wild. Various tools are available, but typically this is accomplished by simply adding your feature to the UseCounter::Feature enum and adding MeasureAs=<your enum value> to the feature's IDL definition.
Fill out the “Motivation” section with a brief summary, and then write up the use cases and scenarios. Proceed to the “Write Up Motivation” stage in ChromeStatus - this will generate an “Intent to Deprecate and Remove” mail for you. Send that email to blink-dev and start checking in your code removing the feature to Chromium as a runtime enabled feature. Make sure to provide suggested alternatives to the feature being deprecated. As soon as you have a functional removal of the feature ready for developers to try out under a flag, proceed to the next step.
Step 2: Dev trial of deprecation
Proceed to the “Dev Trials” stage in ChromeStatus. This will generate a “Ready for Trial” email that you should send to blink-dev to notify the community they can try out the feature deprecation.
At this point, you should also notify developers by adding a deprecation issue as outlined here. Give developers as many milestones as possible to respond to the deprecation.
You should work with the documentation team to ensure the community is prepared for this feature deprecation, and estimate when (in what milestone) you would like to target shipping. You should also decide (possibly based on data from the dev trial) if a deprecation trial is going to be necessary to help smooth the removal of this feature from the web platform.
Step 3 (Optional): Deprecation Trial
If you are concerned that there are going to be web developers who need additional time to fix up their implementations, and will want to delay your feature deprecation, you can file for a Deprecation Trial. This will let you disable the feature by default, but let developers request an origin trial token to re-enable the feature, for a limited period of time after the feature deprecation. You will need to decide how long to keep the deprecation trial open and enter that milestone in the tool for planning purposes, and then select “Draft Request for Deprecation Trial email” in ChromeStatus, and send the resulting “Request for Deprecation Trial” email to blink-dev. After receiving at least one LGTM from the API owners, email firstname.lastname@example.org letting them know you plan to run a deprecation trial, ensure your removal is integrated with the origin trials framework (see details), and Googlers can request a trial for their feature at go/new-origin-trial. Once your Deprecation Trial is in place, proceed to the next step.
Step 4: Prepare to Ship
You should update ChromeStatus with a target milestone for deprecating the feature (and remember to keep this updated, if things change). You should get final signoff from the Documentation team.
Proceed to the “Prepare to Ship” stage in ChromeStatus; this will generate an Intent to Ship mail that you should send to blink-dev. This will spark a conversation with the API owners; address any feedback from them, and once you get 3 LGTMs from the API owners, you may proceed.
Step 5: Disable the feature
Disable the feature by default. Update ChromeStatus to either “Disabled” or “Disabled with Deprecation Trial”.
Step 6: Remove Code
If you are running a Deprecation Trial, wait until the Deprecation Trial period has ended. (If you need to extend the Deprecation Trial, notify email@example.com and click “Generate an Intent to Extend Deprecation Trial” in ChromeStatus and send the resulting notification to blink-dev.)
Once the feature is no longer available, remove the code (typically waiting a couple of milestone cycles to insure the code is no longer needed), and set the ChromeStatus to “Removed.”
If you are unsure of when a feature could be removed, or would like to discourage usage, you may deprecate a feature without a removal deadline. This is strongly discouraged and will require significant justification:
Email blink-dev using the "Intent to Deprecate" template.
1 LGTM necessary from the API owners
Must justify why there is no removal date
* It takes 12-18 weeks to hit Stable once you enable instrumentation. If there is time pressure you may be able to get permission to merge UseCounter changes into an existing dev/beta branch, and make provisional decisions based on data from beta channel.
Web-developer-facing change to existing code (PSA)
Step 1: Write up motivation and implement code
Fill out the “Motivation” section with a brief summary, and proceed to the “Implementing” stage in ChromeStatus.
Step 2: (Optional) Dev trial
If you want to try out this change before shipping it, put your code in Chromium as runtime enabled features, and set the status to “Dev Trial” in ChromeStatus. This will generate a “Ready for Trial” email that you should send to blink-dev to notify the community they can try out code change.
Step 3: Prepare to Ship
You should update ChromeStatus with a target milestone for shipping (and remember to keep this updated, if things change). Proceed to the “Prepare to Ship” stage in ChromeStatus; this will generate a “Web-Facing Change PSA” mail for you. Send that email to blink-dev with the summary of the code change and the expected milestone.
Step 4: (Optional) Finch trial
You may wish to use Finch to increase confidence in the new code as you deploy it.
Step 5: Ship
Ship it, and set the status to “Shipped”.
After launching a new feature, watch for crashes, regressions caused by your feature and any substantive spec feedback. Review UseCounter and other metrics. Update the intent-to-ship thread and ChromeStatus if non-trivial issues (like web compatibility or serious design questions) arise. When in doubt, email blink-dev@ (or blink-api-owners-discuss@ if you prefer a smaller audience) for advice.
Once a new API is on by default, continue to support other implementations (for instance, clarifying the spec, improving the tests, fixing bugs, and updating support status in ChromeStatus) until the feature is broadly supported and works the same across engines. Remember, your job is to move the web forward, not simply add features to Chrome.
Review MDN docs for technical accuracy and clarity. Feel free to make edits directly or reach out to jmedley@.
When you are convinced enabling the feature by default has “stuck” (typically 2 milestones), remove any unused code including RuntimeEnabledFeatures entries.
For an overview, and explanation of motivations, please see:
- Video presentation: Intent to explain: Demystifying the Blink shipping process (Chrome Dev Summit 2019)
- Blog post: Intent to Explain: Demystifying the Blink Shipping Process
- BlinkOn Talk, 2021: Risky Business: Launching Faster In Blink