http/tests/resourceLoadStatistics/aggregate-sorted-data-no-storage-access.html is consistently failing. It started failing in r256011. Results database: https://results.webkit.org/?suite=layout-tests&test=http%2Ftests%2FresourceLoadStatistics%2Faggregate-sorted-data-no-storage-access.html Test run: https://build.webkit.org/builders/Apple-Catalina-Release-WK2-Tests/builds/2771 Test result: https://build.webkit.org/results/Apple-Catalina-Release-WK2-Tests/r256011%20(2771)/results.html Diff: -Registrable domain: subframe3 +Registrable domain: topframe2 hadUserInteraction: No gotLinkDecorationFromPrevalentResource: No - subframeUnderTopFrameDomains: - topframe1 - topframe4 - subresourceUnderTopFrameDomains: - topframe3 - subresourceUniqueRedirectsTo: - topframe2 - isPrevalentResource: Yes + isPrevalentResource: No isVeryPrevalentResource: No dataRecordsRemoved: 0 -Registrable domain: topframe2 +Registrable domain: topframe3 hadUserInteraction: No grandfathered: No @@ -65,12 +58,19 @@ isPrevalentResource: Yes isVeryPrevalentResource: No dataRecordsRemoved: 0 -Registrable domain: topframe3 +Registrable domain: subframe3 gotLinkDecorationFromPrevalentResource: No - isPrevalentResource: No + subframeUnderTopFrameDomains: + topframe1 + topframe4 + subresourceUnderTopFrameDomains: + topframe3 + subresourceUniqueRedirectsTo: + topframe2 + isPrevalentResource: Yes isVeryPrevalentResource: No
<rdar://problem/59259844>
Looks like ResourceLoadStatisticsMemoryStore::dumpResourceLoadStatistics needs to be fixed.
The bug is in existing implementation of ResourceLoadStatisticsMemoryStore::dumpResourceLoadStatistics. It relies on particular hash-table ordering, while hash-table does not guarantee it.
Created attachment 390102 [details] Patch
(In reply to Yusuke Suzuki from comment #3) > The bug is in existing implementation of > ResourceLoadStatisticsMemoryStore::dumpResourceLoadStatistics. It relies on > particular hash-table ordering, while hash-table does not guarantee it. Yes, this patch should fix it. I copied what you did in ResourceLoadStatisticsDatabaseStore.
Comment on attachment 390102 [details] Patch r=me
missed expectations for one test. Uploading a new patch for landing.
Created attachment 390115 [details] Patch for landing
Comment on attachment 390115 [details] Patch for landing Clearing flags on attachment: 390115 Committed r256055: <https://trac.webkit.org/changeset/256055>
All reviewed patches have been landed. Closing bug.