[GTK] http/tests/inspector/network/har/har-page.html is failing

[Miguel Gomez]

http/tests/inspector/network/har/har-page.html [ Failure ]

since it was added in r223856

[Joseph Pecoraro]

What was the expectation diff?

[Michael Catanzaro]

httpVersion is missing:

--- /home/slave/webkitgtk/gtk-linux-64-release-tests/build/layout-test-results/http/tests/inspector/network/har/har-page-expected.txt
+++ /home/slave/webkitgtk/gtk-linux-64-release-tests/build/layout-test-results/http/tests/inspector/network/har/har-page-actual.txt
@@ -29,7 +29,7 @@
         "request": {
           "method": "GET",
           "url": "http://127.0.0.1:8000/inspector/network/har/har-page.html",
-          "httpVersion": "HTTP/1.1",
+          "httpVersion": "",
           "cookies": [],
           "headers": "<filtered>",
           "queryString": [],
@@ -39,7 +39,7 @@
         "response": {
           "status": 200,
           "statusText": "OK",
-          "httpVersion": "HTTP/1.1",
+          "httpVersion": "",
           "cookies": [],
           "headers": "<filtered>",
           "content": {

[Joseph Pecoraro]

Ahh interesting, thanks!

So the `protocol` comes from WebCore::NetworkLoadMetrics, typically populated in the NetworkProcess where NetworkLoadMetrics is constructed. I guess the protocol isn't populated yet by linux, or if it is it has an unexpected value.

For now you could check in a GTK expectation. I don't think this will be flakey.

[Michael Catanzaro]

*** This bug has been marked as a duplicate of bug 179781 ***

[Nikolas Zimmermann]

According to https://build.webkit.org/builders/GTK%20Linux%2064-bit%20Release%20%28Tests%29/builds/11947/steps/layout-test/logs/stdio the test is actually flakey on Gtk.

Changing test expectations.