Invalid frame id doesn't cause a "no such frame" error to be raised
Categories
(Remote Protocol :: Marionette, defect, P1)
Tracking
(firefox-esr60 unaffected, firefox64 wontfix, firefox65 wontfix, firefox66 fixed)
| Tracking | Status | |
|---|---|---|
| firefox-esr60 | --- | unaffected |
| firefox64 | --- | wontfix |
| firefox65 | --- | wontfix |
| firefox66 | --- | fixed |
People
(Reporter: whimboo, Assigned: whimboo)
References
(Blocks 1 open bug, )
Details
Attachments
(3 files, 1 obsolete file)
Originally filed as: https://github.com/mozilla/geckodriver/issues/1456 When accessing the endpoint `session/048b061c-fae2-624c-b701-72c2e9f24b05/frame` and passing in an invalid frame id, we currently raise an invalid argument error. This is most likely a regression from switching to serde (bug 1396821). We should fix it in the 0.24 release of geckodriver.
|
Assignee | |
Comment 1•5 years ago
|
||
This might actually be a bug in the spec. Lets follow up on the Github issue for now until it is clear.
|
|
Assignee | |
Comment 2•5 years ago
|
||
Whatever solution we need, what we clearly miss are wdspec tests, which I will have to write here. Also I noticed a bug in the webdriver rust unit tests, which will be fixed as a ride-along.
|
|
Assignee | |
Comment 3•5 years ago
|
||
We actually will update the WebDriver specification to fix this particular failure. As such lets write some wdspec tests to make sure we cover those invalid data type cases.
|
||
Comment 4•5 years ago
|
||
Sounds great. I hope Google is committed to changing chromedriver
and that this doesn’t cause any user script regression, i.e. that
no one is relying on this behaviour in their tests.
|
|
Assignee | |
Comment 5•5 years ago
|
||
Yes, we are going to change it. As such I filed https://github.com/w3c/webdriver/issues/1387. Sadly my PR was merged too early, which caused two remaining items to be left-over. Those will be fixed with https://github.com/w3c/webdriver/issues/1389.
|
|
Assignee | |
Comment 6•5 years ago
|
||
|
|
Assignee | |
Comment 7•5 years ago
|
||
If the frame id argument of the "Switch To Frame" command is a web element, but not a valid frame object, an "invalid argument" error needs to be thrown instead of a "no such frame" error.
|
|
Assignee | |
Comment 8•5 years ago
|
||
Depends on D15952
|
|
Assignee | |
Comment 9•5 years ago
|
||
Depends on D15953
|
||
Updated•5 years ago
|
|
|
Assignee | |
Comment 10•5 years ago
|
||
|
|
Assignee | |
Updated•5 years ago
|
|
||
Comment 11•5 years ago
|
||
Pushed by hskupin@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/68dbd12b8547 [marionette] Raise "no such element" error in "Switch To Frame" for unknown elements. r=ato https://hg.mozilla.org/integration/autoland/rev/227e15964de8 [geckodriver] Fix unit test for invalid frame id. r=ato https://hg.mozilla.org/integration/autoland/rev/c429964db93c [wdspec] Add tests for "Switch To Frame" command. r=ato
Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/14782 for changes under testing/web-platform/tests
|
||
Comment 13•5 years ago
|
||
| bugherder | ||
https://hg.mozilla.org/mozilla-central/rev/68dbd12b8547
https://hg.mozilla.org/mozilla-central/rev/227e15964de8
https://hg.mozilla.org/mozilla-central/rev/c429964db93c
|
||
Comment 14•5 years ago
|
||
Is this something which requires Beta backport or can it ride the trains?
|
|
Assignee | |
Comment 15•5 years ago
|
||
It's not a regression, and I don't think that we have to uplift it to beta due to the code paths are rarely used.
|
||
Updated•1 year ago
|
Description
•