LayoutTest http/tests/xmlhttprequest/response-empty-arraybuffer.html is failing https://build.webkit.org/results/Apple%20High%20Sierra%20Debug%20WK1%20(Tests)/r222557%20(18)/results.html https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=http%2Ftests%2Fxmlhttprequest%2Fresponse-empty-arraybuffer.html --- /Volumes/Data/slave/highsierra-debug-tests-wk1/build/layout-test-results/http/tests/xmlhttprequest/response-empty-arraybuffer-expected.txt +++ /Volumes/Data/slave/highsierra-debug-tests-wk1/build/layout-test-results/http/tests/xmlhttprequest/response-empty-arraybuffer-actual.txt @@ -8,9 +8,9 @@ PASS Object.prototype.toString.call(request.response) is '[object ArrayBuffer]' PASS request.response.byteLength is 0 Test that XMLHttpRequest.response returns an empty ArrayBuffer when received status is '304 not modified'. -PASS request.status is 200 +FAIL request.status should be 200. Was 206. PASS Object.prototype.toString.call(request.response) is '[object ArrayBuffer]' -PASS request.response.byteLength is 68 +FAIL request.response.byteLength should be 68. Was 3. PASS request2.status is 304 PASS Object.prototype.toString.call(request2.response) is '[object ArrayBuffer]' PASS request2.response.byteLength is 0
This is seen on High Sierra WK1 bots.
Apache doesn't actually return 206 code during this test. So it must be coming from WebKit cache, which seems like a serious WebKit bug (we should check Range header field in request when pulling from cache). 127.0.0.1 - - [27/Sep/2017:11:20:51 -0700] "GET /xmlhttprequest/response-empty-arraybuffer.html HTTP/1.1" 200 2366 570 127.0.0.1 - - [27/Sep/2017:11:20:51 -0700] "GET /xmlhttprequest/resources/zero-length.xml HTTP/1.1" 200 - 508 127.0.0.1 - - [27/Sep/2017:11:20:51 -0700] "GET /xmlhttprequest/resources/reply.xml HTTP/1.1" 304 - 426 Should try reproducing this by running the test after ones that do use range requests with this resource (xmlhttprequest/cache-override.html, xmlhttprequest/range-test.html).
<rdar://problem/34716163>
> seems like a serious WebKit bug If this is truly WebKit1 only, it may be below WebKit in fact.
WebKit cache is not used for WK1. If it is cache related it is an NSURLCache bug.
Confirmed that this is truly WebKit1 only, will keep tracking in Radar.