On Mon, Apr 18, 2022 at 10:49 AM Mike Taylor <miketa...@chromium.org> wrote:
> On 4/12/22 12:08 PM, Domenic Denicola wrote: > > Contact emails > > dome...@chromium.org, jap...@chromium.org > > Explainer > > https://github.com/WICG/navigation-api/blob/main/README.md > > (Aside: This explainer is a master-class in writing explainers. Incredibly > well done - I really appreciate the effort that went into this). > > > Specification > > https://wicg.github.io/navigation-api/ > > Summary > > The window.navigation API provides the ability to intercept and initiate > navigations, as well as introspect an application's history entries. This > provides a more useful alternative to window.history and window.location, > specifically aimed at the needs of single-page web applications. > > (Note: this API was formerly known as the app history API.) > > Blink component > > Blink>History > <https://bugs.chromium.org/p/chromium/issues/list?q=component:Blink%3EHistory> > > TAG review > > https://github.com/w3ctag/design-reviews/issues/605 > > https://github.com/w3ctag/design-reviews/issues/717 > > TAG review status > > Issues addressed > > Link to origin trial feedback summary > > > https://docs.google.com/document/d/1oDtVhNJaDyEAqsthe07wJaGNVpt-g4TLB4A0-lU2lr4/edit?usp=sharing > > Risks > Interoperability > > The biggest interoperability risk with this API is that it is building on > a rocky foundation. The existing session history spec does not match > browsers very well, and browsers do not match each other. Since this new > API layers on top of the existing model, this could cause issues. > > We have attempted to address this via a solid and well-tested > specification for the new API, as well as ongoing efforts in whatwg/html > PR #6315 <https://github.com/whatwg/html/pull/6315> and elsewhere on the > HTML Standard issue tracker > <https://github.com/whatwg/html/issues?q=is%3Aopen+is%3Aissue+label%3A%22topic%3A+navigation%22%2C%22topic%3A+history%22%2C%22topic%3A+browsing+context%22> > to reform the foundational parts of the specification. For example, > although the navigation API's new events, such as currententrychange, are > fired at well-specified times, there is an existing interop problem > <https://github.com/whatwg/html/issues/1792> regarding the timing of > popstate vs. hashchange events, which makes it difficult to write tests for > the ordering of currententrychange vs. hashchange/popstate. Working on such > existing interop issues and specification problems, and then expanding the > navigation API test suite to cover any such interactions, is our team's top > priority after this launch. See also this tracking issue > <https://github.com/WICG/navigation-api/issues/221>. > > I do have slight concerns > <https://github.com/whatwg/html/issues/1792#issuecomment-1101459682> over > the popstate/hashchange event change - I fear that might result in more > back button traps for Chromium users (that sadly Gecko users experience > today). But I could be wrong - do you have any plans to measure and monitor > abuse? Or do we have existing metrics? > To make sure we are on the same page: at this point we are discussing a future Intent to Ship about a separate behavior change, and we are not discussing the Navigation API. Our plan for that future Intent to Ship does indeed involve careful monitoring. However I don't think it has any chance of increasing back-trapping. Deterministically firing the events in the order (sync popstate, async hashchange) like Gecko does, instead of Chrome's version where sometimes it's (async popstate, async hashchange) and sometimes it's (async hashchange, async popstate) depending on network conditions and page size, should not increase back-trapping. > > Regarding whether this new API will be implemented in other browsers, we > have been encouraged by the consistent and positive collaboration with > Gecko engineers, which has led to several API changes and a good amount of > review. (We have no signal from WebKit.) > > Compatibility > > This has been the team's main focus for the last few months, as we burned > through the list of potentially-compat-impacting issues > <https://github.com/WICG/navigation-api/milestone/1>. In collaboration > with Gecko this led to several improvements, such as the API rename (from > app history), a change <https://github.com/WICG/navigation-api/issues/111> > in how replacement navigations are requested, and the addition > <https://github.com/WICG/navigation-api/issues/76> of an indicator for > when a download is requested. We believe the remaining issues (3 at the > time of writing) are manageable: > > > - > > #72 <https://github.com/WICG/navigation-api/issues/72>: this will > result in firing an event more often during extreme edge case scenarios > involving replacement navigations, or in less-rare-but-still-rare scenarios > involving the user clearing their history. Neither case should prove > problematic. > - > > #207 <https://github.com/WICG/navigation-api/issues/207>: the most > likely solution will either be leaving things as they are, or changing the > timing of an event in a way that will not disturb "normal" usage of the > API. Although such a timing change could be risky if this API had wide > deployment, we believe that changing the timing within a milestone or three > would not be problematic if it ends up being desirable. > - > > #202 <https://github.com/WICG/navigation-api/issues/202>: this issue > is about the default for how focus is managed following a navigation > API-intercepted navigation. We believe the currently-chosen default is > probably the best, especially given testaments on that thread from the > accessibility community and from web framework authors. However we have not > yet closed the issue as we haven't concluded the discussion with Gecko > engineers. Similar to #207, this would probably be changeable within a few > milestones if necessary, without significant impact to sites using the API. > And if we did change it, early-adopter sites could easily restore the > previous behavior by changing the value of an option. > > I agree that these issues seem tractable in the near-term. > > > Signals > > Gecko: No signal > <https://github.com/mozilla/standards-positions/issues/543>. Initial > positive opinions on the issue, and continued engagement on the design, but > not yet an official position. > > WebKit: No signal > <https://lists.webkit.org/pipermail/webkit-dev/2021-September/031987.html> > . > > Web developers: Strongly positive > <https://github.com/WICG/proposals/issues/20>. The initial public > proposal, as well as the issue tracker and Twitter, has had great > engagement and enthusiasm from developers. Origin trial feedback was also > encouraging. In addition, we have several conversations going on with > frameworks, libraries, and larger websites to ensure that we're solving the > problems they see with today's history API. So far reactions have been > either positive, or requesting that we add additional functionality (most > notably #32 <https://github.com/WICG/app-history/issues/32>). > > Ergonomics > > Although this API layers onto the same underlying model as window.history, > and will have well-specified interactions with it, the exact integrations > may be confusing. (For example, navigation.navigate() will behave > differently from history.pushState().) We've done our best to smooth over > these rough edges where possible, but have favored making the navigation > API pleasant to use over making it perfectly align with window.history. > > Activation > > This feature is hard to polyfill, but developers have managed to produce > something that works in many cases: frehner/appHistory > <https://github.com/frehner/appHistory> is one, and > virtualstate/navigation <https://github.com/virtualstate/navigation> > another. > > We've also seen a pattern where developers have existing > history/navigation wrappers (e.g. router libraries or app-specific history > and navigation code) which they can adapt with a new navigation API-based > backend for browsers that support it. > > Security > > We believe the security risks of this feature are minimal because of how > it is scoped to same-origin contiguous history entries, and similarly only > allows interception of same-origin navigations. We also need to ensure that > we don't allow "trapping" the user by preventing them from using their back > button; the API is designed to prevent this. > > See the specification's security and privacy discussion > <https://wicg.github.io/navigation-api/#security-privacy> for more. > > WebView Application Risks > > Does this intent deprecate or change behavior of existing APIs, such that > it has potentially high risk for Android WebView-based applications? > > This feature does not introduce any changes to existing APIs. > > Debuggability > > This feature mostly has no need for extended tooling. crbug.com/1252940 > <https://bugs.chromium.org/p/chromium/issues/detail?id=1252940> tracked > adding the newly-introduced events to the Event Listener Breakpoints panel. > > Is this feature fully tested by web-platform-tests > <https://chromium.googlesource.com/chromium/src/+/master/docs/testing/web_platform_tests.md> > ? > > Yes > <https://wpt.fyi/results/navigation-api?label=master&label=experimental&aligned> > . > > These results show a strange number of failures for Chromium. We suspect > this is due to the test runner on wpt.fyi, as running the tests locally, or > in a live Chrome browser, does not exhibit the issue. See > web-platform-tests/wpt#33590 > <https://github.com/web-platform-tests/wpt/issues/33590>. > > Flag name > > NavigationApi > > Requires code in //chrome? > > False > > Tracking bug > > https://bugs.chromium.org/p/chromium/issues/detail?id=1183545 > > Launch bug > > https://bugs.chromium.org/p/chromium/issues/detail?id=1252954 > > Measurement > > https://chromestatus.com/metrics/feature/timeline/popularity/4056 > > Non-OSS dependencies > > Does the feature depend on any code or APIs outside the Chromium open > source repository and its open-source dependencies to function? > > No. > > Sample links > > https://gigantic-honored-octagon.glitch.me/ > > https://selective-heliotrope-dumpling.glitch.me/ > > Estimated milestones > > We plan to land this API in M102. > > Link to entry on the Chrome Platform Status > > https://chromestatus.com/feature/6232287446302720 > > Links to previous Intent discussions > > Intent to Prototype > <https://groups.google.com/a/chromium.org/g/blink-dev/c/R1D5xYccqb0/m/8ukfzdVSAgAJ> > > Intent to Experiment > <https://groups.google.com/a/chromium.org/g/blink-dev/c/ki__L-IiR0Q/m/rG3OgSkKBQAJ> > > > This intent message was generated by Chrome Platform Status > <https://chromestatus.com/> and then cleaned up a good bit. > -- > You received this message because you are subscribed to the Google Groups > "blink-dev" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to blink-dev+unsubscr...@chromium.org. > To view this discussion on the web visit > https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAM0wra8BD%2Bea9fSiRGyPJeAZ2KknQe6c9Xmza5BS7O94ktjXiA%40mail.gmail.com > <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAM0wra8BD%2Bea9fSiRGyPJeAZ2KknQe6c9Xmza5BS7O94ktjXiA%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > > > -- You received this message because you are subscribed to the Google Groups "blink-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to blink-dev+unsubscr...@chromium.org. To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAM0wra9ZG9c34qtOq47V7Z%2B-c%2Bj7NOk_P5RMHhcwHW35fbGXig%40mail.gmail.com.
