1347687 - TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_json.js - TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_rdf.js

Status: RESOLVED FIXED

(Thunderbird :: Installer, enhancement)

Description

[Jorg K (CEST = GMT+2)]

+++ This bug was initially created as a clone of Bug #1347540 +++

TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_json.js | xpcshell return code: 0
TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_rdf.js | xpcshell return code: 0

Alphan and Paolo:
These need tests are failing in the Thunderbird Xpcshell test suite. Please take a look at
https://archive.mozilla.org/pub/thunderbird/tinderbox-builds/comm-central-win32/1489587756/comm-central_win7_ix_test-xpcshell-bm110-tests1-windows-build12.txt.gz
searching for TEST-UNEXPECTED-FAIL. There are plenty of errors logged, more than would make sense to paste here.

How could we get these tests to pass or should we just disable them for Thunderbird.

Comment 1

[Alphan Chen [:alchen]]

Failed log:

(test_handlerService_json.js)
08:53:00     INFO -  TEST-PASS | uriloader/exthandler/tests/unit/test_handlerService_json.js | testIsInHandlerArray - [testIsInHandlerArray : 320] 0 == 0
08:53:00     INFO -  PID 10148 | JavaScript error: resource://gre/components/nsHandlerService-json.js, line 235: NS_ERROR_NOT_AVAILABLE: handlerSvc fillHandlerInfo: don't know this type
08:53:00     INFO -  Unexpected exception NS_ERROR_NOT_AVAILABLE: handlerSvc fillHandlerInfo: don't know this type
08:53:00     INFO -  fillHandlerInfo@resource://gre/components/nsHandlerService-json.js:235:13
08:53:00     INFO -  testIsInHandlerArray@C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests/unit/test_handlerService_json.js -> file:///C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests/unit/common_test_handlerService.js:321:3

(test_handlerService_rdf.js)
08:53:01     INFO -  TEST-PASS | uriloader/exthandler/tests/unit/test_handlerService_rdf.js | testIsInHandlerArray - [testIsInHandlerArray : 320] 0 == 0
08:53:01     INFO -  PID 10268 | JavaScript error: resource://gre/components/nsHandlerService.js, line 324: NS_ERROR_NOT_AVAILABLE: handlerSvc fillHandlerInfo: don't know this type
08:53:01     INFO -  Unexpected exception NS_ERROR_NOT_AVAILABLE: handlerSvc fillHandlerInfo: don't know this type
08:53:01     INFO -  HS_fillHandlerInfo@resource://gre/components/nsHandlerService.js:324:13
08:53:00     INFO -  testIsInHandlerArray@C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests
08:53:01     INFO -  testIsInHandlerArray@C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests/unit/test_handlerService_rdf.js -> file:///C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests/unit/common_test_handlerService.js:321:3

Comment 2

[Alphan Chen [:alchen]]

> file:///C:/slave/test/build/tests/xpcshell/tests/uriloader/exthandler/tests/
> unit/common_test_handlerService.js:321:3

@common_test_handlerService.js:321
gHandlerService.fillHandlerInfo(protoInfo, "ircs");


Two failed cases are the same.
When doing fillHandlerInfo with overrideType "ircs", the handler service didn't know this type.

In Firefox, irc handler info will be injected by injectNewDefaults().
http://searchfox.org/mozilla-central/source/uriloader/exthandler/nsHandlerService.js#143
http://searchfox.org/mozilla-central/source/uriloader/exthandler/nsHandlerService-json.js#68

If Thunderbird doesn't do the injection, I think you can just pass this test as what we did on Fennec.
http://searchfox.org/mozilla-central/source/uriloader/exthandler/tests/unit/common_test_handlerService.js#312

Comment 3

[Alphan Chen [:alchen]]

There is another way to solve this.

Since you can pass the test "testReloadWithoutDB", I think you can also use the preference "defaultHandlersVersion" to do the check at the beginning of the test "testIsInHandlerArray".
You can add a check and do an early return if the preference does not exist.

  if (!Services.prefs.getPrefType("gecko.handlerService.defaultHandlersVersion")){
    return;
  }

Comment 4

[Jorg K (CEST = GMT+2)]

Thank you for your answer. How about just adding those preferences to TB?
pref("gecko.handlerService.defaultHandlersVersion", "chrome://browser-region/locale/region.properties");
and many more.

Let's try this here:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=3e79665f518dfb1c8b4f672fb6d9dccbed30c2ec

Comment 5

[Jorg K (CEST = GMT+2)]

That made it even worse:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&selectedJob=84234986

https://archive.mozilla.org/pub/thunderbird/try-builds/[e-mail]-3e79665f518dfb1c8b4f672fb6d9dccbed30c2ec/try-comm-central-macosx64/try-comm-central_yosemite_r7_test-xpcshell-bm134-tests1-macosx-build7.txt.gz

So next I'll try disabling according to comment #3.

Comment 6

[Jorg K (CEST = GMT+2)]

https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&revision=925ebfcc7eff691a2ee23d816d99280427f882fb

Comment 7

[Alphan Chen [:alchen]]

(In reply to Jorg K (GMT+1) from comment #4)
> Thank you for your answer. How about just adding those preferences to TB?
> pref("gecko.handlerService.defaultHandlersVersion",
> "chrome://browser-region/locale/region.properties");
> and many more.
> 
If you want to injectDefault as Firefox, you need to add more preferences.
"gecko.handlerService.defaultHandlersVersion" is not the only one.
You can check the following for more detail.
http://searchfox.org/mozilla-central/source/browser/app/profile/firefox.js#731

Comment 8

[Jorg K (CEST = GMT+2)]

I tried two things:

First, I added a lot of preferences and got a failure, see comment #5:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&selectedJob=84234986
Sorry about all the other failures, but you can see that test_handlerService.js and friends still failed.
Here are all the preferences I added, which one did I miss?
https://hg.mozilla.org/try-comm-central/rev/d713d1e9d1eedb90dd68c303c6915fe749d9da31

In the next try I did what you had suggested in comment #3, that stopped the failures, so let's go with this.

Comment 9

[Jorg K (CEST = GMT+2)]

Attachment: 1347687-skip-test.patch

Solution suggested in comment #3. Or should I replace the
   if (Services.appinfo.widgetToolkit == "android") {
     do_print("Skipping test because it does not apply to this platform.");
     return;
   }
with the new test?

Comment 10

[Jorg K (CEST = GMT+2)]

Comment on attachment 8848076 [details] [diff] [review]
1347687-skip-test.patch

Sorry, I didn't see that this failure *hadn't* gone away since there were to many other failures in that try run. Looking again here:
https://treeherder.mozilla.org/#/jobs?repo=try-comm-central&selectedJob=84249183
we still have:

TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_json.js | xpcshell return code: 0
TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_json.js | testFillHandlerInfoForProtocol - [testFillHandlerInfoForProtocol : 377] false == true
TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_rdf.js | xpcshell return code: 0
TEST-UNEXPECTED-FAIL | uriloader/exthandler/tests/unit/test_handlerService_rdf.js | testFillHandlerInfoForProtocol - [testFillHandlerInfoForProtocol : 377] false == true

That's with the additional
+  if (!Services.prefs.getPrefType("gecko.handlerService.defaultHandlersVersion")) {
+    do_print("Skipping test: No pref gecko.handlerService.defaultHandlersVersion.");
+    return;
+  }
in testIsInHandlerArray().

Comment 11

[Jorg K (CEST = GMT+2)]

Attachment: 1347687-skip-test.patch (v2)

Sorry, skipping the test needed to be added twice. Now the test "passes" locally.

Comment 12

[Alphan Chen [:alchen]]

Comment on attachment 8848097 [details] [diff] [review]
1347687-skip-test.patch (v2)

Review of attachment 8848097 [details] [diff] [review]:
-----------------------------------------------------------------

LGTM

Comment 13

[Jorg K (CEST = GMT+2)]

Hi Tomcat, this needs to be landed on M-C/M-I. I didn't want to move the bug to an M-C product/component, so I'm setting NI so you can pick it up. Thank you in advance.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/d913a0c7d25a
Skip test if pref gecko.handlerService.defaultHandlersVersion not defined. r=alchen

Comment 15

[Carsten Book [:Tomcat]]

landed :)

Comment 16

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/d913a0c7d25a