LGTM1
Thanks for the extra details. These results look amazing, so in my
mind they significantly outweigh the risk of going first here.
On Thursday, May 12, 2022 at 4:12:19 AM UTC+2 Hiroki Nakagawa wrote:
Let me share more details of the feedback.
We worked directly with 11 external partners; 8 of them valued the
opportunity offered by Prerendering.
* Origin Trial results: One of the external partners, Ameba News
(https://news.ameba.jp/) has tested out same-origin
prerendering via the Origin Trial during April 11 - 17 (1 week
testing / around selected navigations / 2-3% of the traffic).
While the site was already meeting CWV and good LCP, by
adopting prerendering, the median LCP improved 2.7x (827ms to
302ms) and more importantly the P95 improved by 1.6 seconds
[2843ms to 1195ms] (i.e. Prerendered pages have more
consistency in the performance).
* Key challenges:
o All participants identified "knowing what is safe to
prerender / adapting a page to be safe" as a key
challenge. We are exploring follow-ups to lower the bar.
o DevTools support. We are following up with the devtools team.
o 3P RUM tools to incorporate `activationStart` to correctly
measure FCP / LCP. We will be guiding the RUMs for
required changes.
o CrUX to incorporate the prerendering impact. We are
following up with the CrUX team.
In addition, we worked with one internal partner who got great
results (i.e. the 45% number shared earlier).
Finally, we received the following feedback from 6 OT participants:
* Difficulty: moderately easy (3), neither easy nor difficult
(2), Slightly difficult (1)
* Keep using: extremely likely (2), moderately likely (3), can't
say which way (1)
* Alternatives: none (2), fallback to prefetch (2)
Thanks,
Hiroki
On Tue, May 10, 2022 at 10:42 PM Hiroki Nakagawa
<[email protected]> wrote:
Hi Yoav, thanks for your comments!
On Tue, May 10, 2022 at 9:48 PM Yoav Weiss
<[email protected]> wrote:
On Fri, Apr 22, 2022 at 7:27 PM 'Angel Raposo' via
blink-dev <[email protected]> wrote:
Contact emails
[email protected][email protected],
[email protected], [email protected],
[email protected]
Explainer
This feature:
https://github.com/WICG/nav-speculation/blob/main/same-origin-explainer.md
<https://github.com/WICG/nav-speculation/blob/main/same-origin-explainer.md>
Larger project:
https://github.com/WICG/nav-speculation/blob/main/README.md
<https://github.com/WICG/nav-speculation/blob/main/README.md>
Specification
https://wicg.github.io/nav-speculation/prerendering.html
<https://wicg.github.io/nav-speculation/prerendering.html>
Design docs
https://docs.google.com/document/d/1P2VKCLpmnNm_cRAjUeE-bqLL0bslL_zKqiNeCzNom_w/edit?usp=sharing
<https://docs.google.com/document/d/1P2VKCLpmnNm_cRAjUeE-bqLL0bslL_zKqiNeCzNom_w/edit?usp=sharing>
Summary
Prerendering is “pre”-rendering, it’s about
pre-loading and rendering a Web page before the user
actually navigates to it. The main goal of
prerendering is to make the next page navigation
faster, or ideally nearly instant.
Sites can tell the user agents about which pages the
user may likely visit, by asking to trigger a
‘prerendering’ for a particular URL (e.g. user is at
page A and will likely navigate to page B next). Once
the prerender is triggered, the browser pre-fetches
the main resource, instantiates a hidden page, and
processes the main resource to fetch and process more
subresources.
We would like to ship same-origin speculation rules
prerendering. With this feature, Chrome will start
prerendering high-confidence URL suggestions provided
by the page using speculation rules. During the
prerendering process, a page will process and
construct the full DOM tree, including the execution
of scripts (this differs from the prefetching
resources using No-state Prefetch
<https://developers.google.com/web/updates/2018/07/nostate-prefetch>)
This prerendering affects how the prerendered pages
are processed. The website gets loaded before the
navigation is committed, and the side-effects are
delayed until activation (the actual navigation to the
website was committed). The website can know that it
is being prerendered (document.prerendering) and when
it was activated (prerenderingchange event,
performance.activationStart timing). These APIs have
already been approved by the previous intent-to-ship
for Omnibox prerendering
<https://docs.google.com/document/d/1GcB_oII6-3s24M_3ebcp5bzAtNNY5i_wh1c0jgJvv8k/edit>.
Note that we are not shipping cross-origin speculation
rules, which allows a web page to prerender another
page under a different domain (external site). For
details about our web exposed changes shipping plan,
please consult this doc: Interoperability Roadmap for
Shipping Prerender2 Incrementally
<https://docs.google.com/document/d/1GenZDPbTzRyp038gh8A_ylIoVTfKt40MFyg7Ek_7L58/edit>.
Blink component
Internals>Preload>Prerender
<https://bugs.chromium.org/p/chromium/issues/list?q=component:Internals%3EPreload%3EPrerender>
TAG review
https://github.com/w3ctag/design-reviews/issues/667
<https://github.com/w3ctag/design-reviews/issues/667>
TAG review status
Closed
Risks
Interoperability and Compatibility
https://github.com/WICG/nav-speculation/issues/160
<https://github.com/WICG/nav-speculation/issues/160> seems
like something we'd want to settle before shipping?
I wonder if the issue is not blocking this I2S. Currently we
support only the basic syntax like this
<https://github.com/web-platform-tests/wpt/blob/88cd60961f5cd877cf5083a8874a233f69ae72a2/speculation-rules/prerender/resources/utils.js#L19-L22>,
and such advanced syntax is out of the scope. I think we can
incrementally support them later without breaking the
compatibility.
Any other open issues that may impact the compat or
interop risk here?
We audited open issues that could potentially affect compat or
interop and concluded they have no risks. This document
<https://docs.google.com/document/d/1eFY7RMoeG7Mdhon9yLs6hKSfi6DYrASBPM-31hWXPDg/edit?usp=sharing>
summarizes the results.
Also, is the opt-in
<https://github.com/WICG/nav-speculation/blob/main/opt-in.md>
part of this intent?
This I2S doesn't cover the opt-in mechanism.
Compatibility
There are some use cases that will need to know
whether a page is being prerendered by the user agent
or navigated by the user, e.g. Ads and analytics are
likely examples of this.To solve this challenge, this
feature exposes `document.prerendering` which
indicates if the page is prerendered or not. We
understand that there’s always a risk of sites, that
would benefit from using this API, not actually using it.
We believe that this risk is tractable because:
(1) triggering a prerender isn’t an entirely new
behavior. E.g. Chrome used to have a prerendering
feature a long while ago
<https://blog.chromium.org/2011/06/prerendering-in-chrome.html>triggered
by URLs entered in the navigation bar
(2) Similar behaviors are already present in other
browsers, e.g. “Preload Top Hit” feature in Safari
To give more flexibility to web authors (site owners)
who want to implement speculation rules, we are
already including a header to network requests as
`Purpose: prefetch` so that origin servers can
identify requests for prerendered pages and decide how
to handle them. A new header will be included to
provide a CORS safelist alternative using
`Sec-purpose: prefetch;prerender`
https://fetch.spec.whatwg.org/#cors-safelisted-request-header
<https://fetch.spec.whatwg.org/#cors-safelisted-request-header>
https://chromestatus.com/feature/6247959677108224
<https://chromestatus.com/feature/6247959677108224>
There’s also a future risk of the speculation rules
specification changing
<https://github.com/WICG/nav-speculation/blob/main/triggers.md#rules>:
including new rules, modifying previous ones, etc. We
do not expect to make backward-incompatible changes to
the speculation rules format, but if we did, the user
agent wouldn’t be able to understand the rules (which
would define them as null), opting out of the
prerender which wouldn’t cause any visible effect
(apart from a slower loading time for the next
navigated page as it won’t be prerendered).
We are working closely on the tracker
https://github.com/WICG/nav-speculation/issues
<https://github.com/WICG/nav-speculation/issues>which
contains currently 8 non-blocking open discussions
related with compatibility (1 of them covering
cross-origin which we are not shipping with this I2S)
Interoperability
We believe that some browsers already have
prerendering implementations which are not
well-specified and may differ from each other. Our
vision is to produce a specification that can help
improve interoperability. There is a risk that other
browsers do not converge on a prerendering standard.
The danger here is that different browsers have
different ways to trigger a prerendered page, and
prerendered pages behave differently in different
browsers.
Prerendering is a web-visible behavior, since it
involves fetching the page and executing its scripts.
Prerendering can depend on UA-specific heuristics. For
example, the browser might decide to act on a hint to
prerender based on the system load, and the presence
of other prerenders. We do not intend to codify
heuristics in the specification. A conforming browser
might simply ignore all hints to prerender a page.
Discussions around speculation rules and the broader
prerender feature has already been initiated with
different players:
Gecko: Some informal positive discussion with Gecko
engineers on the HTML Standard issue tracker
<https://github.com/whatwg/html/issues/7533#issuecomment-1022051187>and
in the HTML triage call
<https://github.com/whatwg/html/issues/7488#issuecomment-1029510684>;
formal positions request here:
https://github.com/mozilla/standards-positions/issues/613
<https://github.com/mozilla/standards-positions/issues/613>
WebKit: WebKit already ships URL-bar triggered
prerendering, but not any APIs for letting pages know
about it, and it's unclear what strategy they are
using to prohibit disruptive behaviors for prerendered
pages. We have reached out for a formal positions
request here in the hopes of moving toward
interoperability:
https://lists.webkit.org/pipermail/webkit-dev/2022-February/032113.html
<https://lists.webkit.org/pipermail/webkit-dev/2022-February/032113.html>
Web developers: We’ve already received positive
feedback from initial web developers testing the
feature. See
https://github.com/WICG/proposals/issues/2
<https://github.com/WICG/proposals/issues/2>for
positive sentiments on speculation rules triggered by
different partners who are willing to implement the
feature into their sites.
It'd be great if you could share more details about the
origin trial, feedback we received and if possible, user
benefits from OT participants trying this out.
We had experiments with several internal/external partners.
One of the experiments showed that the feature can reduce the
latency on their service by 45%. Other experiments also showed
promising impacts on the CWV metrics. According to feedback
from the partners, the lack of DevTools support for
prerendering was challenging. As the "Debuggability" section
on this I2S mentions, we've actively been developing the
DevTools support, so the situation should be mitigated soon.
Overall, they were supportive of the feature and interested in
using it on production.
Other signals: There’s a public request for feedback
published
(https://web.dev/speculative-prerendering/#feedback-welcome
<https://web.dev/speculative-prerendering/#feedback-welcome>)
which is currently being managed through:
https://github.com/WICG/nav-speculation/issues
<https://github.com/WICG/nav-speculation/issues>
Ergonomics
IT admins can disable Prerendering via the existing
group policy "NetworkPredictionOptions"
This feature will be used in combination with the APIs
which already shipped as part of omnibox prerendering:
document.prerendering, prerenderingchange event, and
performanceEntry.activationStart timing.
Activation
Developer needs to opt-in by defining the speculation
rules on the page. Please notice that this doesn’t
ensure the prerender of the next navigation as
speculation rules are only used as suggestions to
improve the heuristics of the user agent choosing
which page to prerender. As hints, speculation rules
may be followed by the browser (prerendering the page)
or ignored completely (e.g. lowend device without
resources to prerender the page).
Security
This feature is the first use of the Multiple-Page
Architecture, which is a significant change to
Chromium's internals. Both MPArch and this feature in
particular underwent significant security review. See
the design doc for more details.From a web-exposed
perspective, the security and privacy concerns are
smaller, because this feature is restricted to the
same-origin case only.
WebView application risks
This feature doesn’t deprecate or change behavior of
existing APIs directly.
Debuggability
We are actively talking to the DevTools team about
adding general Prerender support to it [metabug
<https://bugs.chromium.org/p/chromium/issues/detail?id=1217029>].
The current MVP is to reveal the prerendered page
status in the panel so web developers can know if they
finished successfully or not.
We also have a long term plan
<https://docs.google.com/document/d/1nK1TqDMeE0dgl4TGk9Ne4vzKzgv7nkfgmAVbvR6TtM0/edit>to
show different debugging status messages in DevTools
but there's no specific messages currently discussed
that are focused only on speculation rules, only the
"generic" ones at prerender feature level. Your
thoughts are welcome if specific debugging logs
regarding speculation rules would be required.
Is this feature fully tested by
web-platform-tests
<https://chromium.googlesource.com/chromium/src/+/master/docs/testing/web_platform_tests.md>?
We have written a lot of web platform tests using
speculation rules as the trigger:
https://wpt.fyi/results/speculation-rules/prerender?label=experimental&label=master&aligned
<https://wpt.fyi/results/speculation-rules/prerender?label=experimental&label=master&aligned>
(Currently Chromium tests this using VirtualTestSuites
[cs
<https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/web_tests/virtual/prerender/README.md>]
instead of EnableExperimentalWebPlatformFeatures, so
the tests are not green on wpt.fyi.)
Flag name
Prerender2
Requires code in //chrome?
False
Tracking bug
https://crbug.com/1295143 <https://crbug.com/1295143>
Launch bug
https://crbug.com/1316557 <https://crbug.com/1316557>
Sample links
For a demo of prerendering generally, and its effect
on websites, https://prerender2-specrules.glitch.me/
<https://prerender2-specrules.glitch.me/>shows it
triggered by speculation rules.
Estimated milestones
OriginTrial android last
102
OriginTrial android first
94
DevTrial on android
94
Anticipated spec changes
There are currently 8 open discussions
<https://github.com/WICG/nav-speculation/issues?q=is%3Aissue+is%3Aopen+label%3A%22prerendering%22+label%3A%22affects-compat%22+>which
could potentially modify the API in the future. Having
said that, half of them were opened in 2020/2021 which
means they have been in discussions for a long term
and we don’t expect immediate changes required in the
short term.
We have reviewed all the current pending discussions
<https://docs.google.com/document/d/1eFY7RMoeG7Mdhon9yLs6hKSfi6DYrASBPM-31hWXPDg/edit>and
we believe they won’t cause compatibility issues as
most of them will be resolved with the initial launch.
We’ll keep working on the tracker and updating our
roadmap doc
<https://docs.google.com/document/d/1GenZDPbTzRyp038gh8A_ylIoVTfKt40MFyg7Ek_7L58/edit?usp=sharing>if
we detect any changes.
Link to entry on the Chrome Platform Status
https://chromestatus.com/feature/5355965538893824
<https://chromestatus.com/feature/5355965538893824>
Links to previous Intent discussions
Intent to Experiment:
https://groups.google.com/a/chromium.org/g/blink-dev/c/3JwGNnqH3QI/m/Ey103Q-yAQAJ
<https://groups.google.com/a/chromium.org/g/blink-dev/c/3JwGNnqH3QI/m/Ey103Q-yAQAJ>
Intent to Extend Experiment:
https://groups.google.com/a/chromium.org/g/blink-dev/c/Kpp6uJJRrqI/m/GTo_aF0qEQAJ
<https://groups.google.com/a/chromium.org/g/blink-dev/c/Kpp6uJJRrqI/m/GTo_aF0qEQAJ>
This intent message was generated by Chrome Platform
Status <https://chromestatus.com/>.
--
Angel Raposo |TPM Prerender (20% role)
|[email protected] | Google Japan G.K.
This email may be confidential or privileged. If you
received this communication by mistake, please don't
forward it to anyone else, please erase all copies and
attachments, and please let me know that it went to
the wrong person. Thanks.
The above terms reflect a potential business
arrangement, are provided solely as a basis for
further discussion, and are not intended to be and do
not constitute a legally binding obligation. No
legally binding obligations will be created, implied,
or inferred until an agreement in final form is
executed in writing by all parties involved.
もし、このメッセージが誤って貴殿に送信されたと思われる場合には、機密情報を含んでいる可能性もありますので、どなたにも転送せず、添付ファイルも含めて削除していただくとともに、発信者にその旨をお伝えいただきますようお願いいたします。
--
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
[email protected]
<mailto:[email protected]>.
To view this discussion on the web visit
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAA9vRHxK26yKNbkcL3uWGcGHGDEZpQBSBy3Avbp%2BqAV4iTpaOg%40mail.gmail.com
<https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAA9vRHxK26yKNbkcL3uWGcGHGDEZpQBSBy3Avbp%2BqAV4iTpaOg%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 [email protected].
To view this discussion on the web visit
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/95894ef5-83cb-4906-a8c8-fad0b0e71f4fn%40chromium.org
<https://groups.google.com/a/chromium.org/d/msgid/blink-dev/95894ef5-83cb-4906-a8c8-fad0b0e71f4fn%40chromium.org?utm_medium=email&utm_source=footer>.
