Consider reducing the ad attribution data budget to 3 bits #28
Assignees
Labels
layering
Layering additional data and functionality on top of PCM
Comments
|
Ping @michaelkleber @csharrison. |
The discussion on #11 indicated support for a 4+4+4 split (so potentially 8 and 4). I would welcome that. In practice, the 64 ad campaigns limit will be more problematic for us than the reduction in the fidelity of the conversion value. |
|
Agreed with @benjaminsavage. Larger domains will have a lot of campaigns, especially the marketplaces like Target or Shopify that have lots of uncoordinated advertisers running their own campaigns. My preference would be 9-3 over 6-6 because that would give us ~500 campaigns per domain and also align the conversion metadata spec with the Chrome proposal. |
5 tasks
|
@johnwilander Hi! For planning purposes, it would be helpful to understand where we ended up here. Thanks! |
|
I landed 8+4 bits for PCM in WebKit today: https://trac.webkit.org/changeset/270456 That is, 8 bits for the source ID (previously campaign ID) and 4 bits for the trigger data (previously conversion value). |
caiolima
pushed a commit
to caiolima/webkit
that referenced
this issue
Dec 7, 2020
https://bugs.webkit.org/show_bug.cgi?id=219519 <rdar://problem/70470036> Reviewed by Brent Fulgham. We've received a lot of feedback saying increased entropy on the click side is more important than the current 6 bits on the conversion side. Some of that conversation is captured in privacycg/private-click-measurement#28. Source/WebCore: This patch switches from 6+6 bits to 8+4 bits. It also fixes some minor logging issues. Existing layout tests and API tests were updated. * html/HTMLAnchorElement.cpp: (WebCore::HTMLAnchorElement::parsePrivateClickMeasurement const): * loader/PrivateClickMeasurement.cpp: (WebCore::PrivateClickMeasurement::parseAttributionRequest): Removed the check that would log "Conversion was not accepted because the URL path did not start with ..." on every redirect in PCM Debug Mode. It was wrong and annoying. (WebCore::PrivateClickMeasurement::json const): Added a call to isValid(). Other checks made sure this wasn't an issue but I'd rather have it in this public function too. * loader/PrivateClickMeasurement.h: (WebCore::PrivateClickMeasurement::AttributionTriggerData::isValid const): Source/WebKit: This patch switches from 6+6 bits to 8+4 bits. It also makes sure PCM Debug Mode consistently logs on the LOG level except for real errors. Existing layout tests and API tests were updated. * NetworkProcess/PrivateClickMeasurementManager.cpp: (WebKit::PrivateClickMeasurementManager::storeUnattributed): (WebKit::PrivateClickMeasurementManager::attribute): (WebKit::PrivateClickMeasurementManager::fireConversionRequest): Tools: This patch switches from 6+6 bits to 8+4 bits. * TestWebKitAPI/Tests/WebCore/PrivateClickMeasurement.cpp: (TestWebKitAPI::TEST): LayoutTests: This patch switches from 6+6 bits to 8+4 bits. * http/tests/privateClickMeasurement/anchor-tag-attributes-validation-expected.txt: * http/tests/privateClickMeasurement/anchor-tag-attributes-validation.html: git-svn-id: http://svn.webkit.org/repository/webkit/trunk@270456 268f45cc-cd09-0410-ab3c-d52691b4dbfc
ryanhaddad
pushed a commit
to WebKit/WebKit
that referenced
this issue
Dec 22, 2020
https://bugs.webkit.org/show_bug.cgi?id=219519 <rdar://problem/70470036> Reviewed by Brent Fulgham. We've received a lot of feedback saying increased entropy on the click side is more important than the current 6 bits on the conversion side. Some of that conversation is captured in privacycg/private-click-measurement#28. Source/WebCore: This patch switches from 6+6 bits to 8+4 bits. It also fixes some minor logging issues. Existing layout tests and API tests were updated. * html/HTMLAnchorElement.cpp: (WebCore::HTMLAnchorElement::parsePrivateClickMeasurement const): * loader/PrivateClickMeasurement.cpp: (WebCore::PrivateClickMeasurement::parseAttributionRequest): Removed the check that would log "Conversion was not accepted because the URL path did not start with ..." on every redirect in PCM Debug Mode. It was wrong and annoying. (WebCore::PrivateClickMeasurement::json const): Added a call to isValid(). Other checks made sure this wasn't an issue but I'd rather have it in this public function too. * loader/PrivateClickMeasurement.h: (WebCore::PrivateClickMeasurement::AttributionTriggerData::isValid const): Source/WebKit: This patch switches from 6+6 bits to 8+4 bits. It also makes sure PCM Debug Mode consistently logs on the LOG level except for real errors. Existing layout tests and API tests were updated. * NetworkProcess/PrivateClickMeasurementManager.cpp: (WebKit::PrivateClickMeasurementManager::storeUnattributed): (WebKit::PrivateClickMeasurementManager::attribute): (WebKit::PrivateClickMeasurementManager::fireConversionRequest): Tools: This patch switches from 6+6 bits to 8+4 bits. * TestWebKitAPI/Tests/WebCore/PrivateClickMeasurement.cpp: (TestWebKitAPI::TEST): LayoutTests: This patch switches from 6+6 bits to 8+4 bits. * http/tests/privateClickMeasurement/anchor-tag-attributes-validation-expected.txt: * http/tests/privateClickMeasurement/anchor-tag-attributes-validation.html: Canonical link: https://commits.webkit.org/232134@main git-svn-id: https://svn.webkit.org/repository/webkit/trunk@270456 268f45cc-cd09-0410-ab3c-d52691b4dbfc
|
This has now been updated in the spec too: f7e51be |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
(And bumping the campaign ID budget up accordingly.)
PCM proposes a 6-bit value encoding the conversion on the click destination side. The Click Through Conversion Measurement Event-Level API proposal has a corresponding 3-bit value called
conversion-metadata. Ideally, we’d reconcile this into something simple for developers.To discuss:
The text was updated successfully, but these errors were encountered: