Do note that this test is disabeld in the manifest.
r=dburns
--HG--
extra : commitid : HvrxGUF6QhD
extra : rebase_source : 370886c6a60b132ac9406c38ca42fce78301190f
In order to meet the addon signing requirement for tests, specialpowers
needs to be installed at gecko runtime. This means it must be restartless.
This patch packages specialpowers as a restartless addon, but it does not
yet install it at runtime.
--HG--
rename : testing/specialpowers/components/SpecialPowersObserver.js => testing/specialpowers/content/SpecialPowersObserver.jsm
extra : commitid : Bbg8gFten8S
extra : rebase_source : 99d0b841a1f8f9eb6b6ca846363a158836482e57
Test AccessibleCaretEventHub is properly initialized if <html> has
"display: none" initially.
--HG--
extra : commitid : 4D5733d5Xr8
extra : rebase_source : 9c767b200a63777b3f20e85ebc3f83272e979bc9
Services.appinfo is populated by the target compile system's OS name,
taken from the OS_TARGET configure variable.
By using Services.sysinfo instead, we are returning the effective current
local system's operating system, which is similar to what is found in
Gecko's UA string.
r=dburns
--HG--
extra : commitid : 2ap54ga0CLq
extra : rebase_source : 803baa9d40dee8630e58b8ccf6d68ef7ca6f0e3b
This is needed to install unsigned addons in developer mode. A special pref that
can only be set at runtime will determine whether or not unsigned addons can be
installed. So unsigned addons required for testing will need to be restartless
and installed on the fly. The normal method of dropping the addon in the profile
folder will no longer work.
--HG--
extra : commitid : HaXy0GQQmz9
extra : rebase_source : 4ba87c3d40a7fea9dd7afc510ce16ab54019e32b
The Python client does not currently exhaust all the command
requests originating from the server as pointed out by :bhsu in
https://bugzilla.mozilla.org/show_bug.cgi?id=1211503#c12.
This prevents making multiple calls to runEmulatorCmd and runEmulatorShell
inside executeScript/executeJSScript/executeAsyncScript. This failure
makes marionette-webapi fail.
We loop through all the commands originating from the server until we
get sent back a response.
r=dburns
--HG--
extra : commitid : 1SVHwcJrDcU
The default behaviour is unchanged: gecko.log and logcat are saved to the
cwd, a temporary Firefox profile is cloned or created in TMP.
When a path is specified via --workspace, gecko.log, logcat and the Firefox
profile directory are all saved there instead. A profile saved in the
workspace does not get deleted at the end of a session unless it's a clone.
--HG--
extra : rebase_source : 155c26af065f55bcba0f2a139c9f06e4a526f740
The Python client does not currently exhaust all the command
requests originating from the server as pointed out by :bhsu in
https://bugzilla.mozilla.org/show_bug.cgi?id=1211503#c12.
This prevents making multiple calls to runEmulatorCmd and runEmulatorShell
inside executeScript/executeJSScript/executeAsyncScript. This failure
makes marionette-webapi fail.
We loop through all the commands originating from the server until we
get sent back a response.
r=dburns
--HG--
extra : commitid : 8oBRRddi4TJ
extra : rebase_source : 98b10ccaba525acd391bf614a91aec1b8f7daf4e
Services.appinfo is populated by the target compile system's OS name,
taken from the OS_TARGET configure variable.
By using Services.sysinfo instead, we are returning the effective current
local system's operating system, which is similar to what is found in
Gecko's UA string.
r=dburns
--HG--
extra : commitid : 2UVeU9zzO6p
extra : rebase_source : cea1d6f8bc00173a68a28b51df33d59320be8093
Corrects type checks on parameters passed to command, indentation level,
and clarifies when the code leaps into content space.
Thanks to Stanislas Daniel Claude Dolcini for doing the first iteration
fix for this.
r=dburns
--HG--
extra : commitid : 3HeTTEsvvBx
extra : rebase_source : 7088760ae24d695a2f9fa20b14a16cdec2e125d7
There have been a number of improvements to the marionette test runner
unfortunately we missed this regression.
--HG--
extra : commitid : KE4FPolRysN
extra : rebase_source : 546152a67a4fd96a1d3c63ba914839c316022eaf
* BaseMarionetteRunner: any change to self.bin resets self.marionette and
self.tests, so you change binaries between calls to run_tests()
* Refactor runtests.py to make it easier to extend/customize the harness
for different test suites. Add more error handling.
--HG--
extra : transplant_source : %BB%AC%A0%93%A1s%7F%23%A5%DER%E0%AF%B8Q%AF%83%98d%D9
Prepare release of new Python packages for Marionette, with the following
versions:
marionette-client 1.1.0
marionette-driver 1.1.0
marionette-transport 1.0.0
r=dburns
--HG--
extra : commitid : DelEI4Q7Gnu
extra : rebase_source : 7cf900d42fbd60e44e6855c93807a7ba1409fb16
This allows us from a child frame to go up one level. This is documented
at http://w3c.github.io/webdriver/webdriver-spec.html#switch-to-parent-frame
--HG--
extra : commitid : GtXAxWLM61v
extra : rebase_source : 9b18d51a5dbefb4dad1cb39b229dc3d55cfae26a
extra : histedit_source : d0a0479e0c9946f242efd71e722685447ae48fd1
Futures-proofs the Marionette Python client to have support for level
3 of the Marionette protocol outlined in bug 1211489.
This patch changes the marionette-transport API most notably by renaming
the MarionetteTransport class to TcpTransport and by splitting the
receive-capabilities of TcpTransport.send into a new function called
request.
Furthermore it introduces a message data structure for dealing with
incoming responses and commands, and for marshaling messages to send in
order to support all three protocol levels.
r=dburns
r=jgriffin
--HG--
extra : commitid : 3jwoFepZzgy
extra : rebase_source : 5012e774b6edf6a1d53db22e360b5a0b063dd59b
We already had focus changing tests covered by
_test_focus_obtained_by_long_press(). This patch adds tests for focus
not being changed from an editable element after long-pressing on a
non-selectable element.
These tests will fail after reverting this change.
"Bug 1197739 - Do not change focus too early unless the frame is selectable"
https://hg.mozilla.org/mozilla-central/rev/4f2822bbbdb2
--HG--
extra : transplant_source : %26%E8%BAx%D2y%3DS%5BvLH%0Fy%11N%2B%D9%3B%E0
Errors thrown by takeScreenshot used to be silently ignored. When the
command started using the new dispatching technique in bug 1202663,
it was surfaced we do not support taking screen captures of SVG documents.
Since this is a requirement for Web Platform Tests, this patch corrects
the wrong assumptions about document body and document element.
This patch also significantly refactors the screen capture code, but
only uses the new implementation in contnent space, since some further
modifications are required to use it in chrome.
r=dburns
r=jgriffin
--HG--
extra : commitid : DdCIEpd5PEJ
extra : rebase_source : 7357010f992d7f995765c685000892cc59d9ec9a
The screen capture tests were not testing that the screenshots are of
what they're meant to be of. In order to confidently fix bug 1202663,
these new tests will ensure we do not have any regressions.
r=jgriffin
r=dburns
--HG--
extra : commitid : AulnylZMj6Y
extra : rebase_source : c2604e0e3bf1a975b1f35a149ef7f2bdb08840de
The screen capture tests were not testing that the screenshots are of
what they're meant to be of. In order to confidently fix bug 1202663,
these new tests will ensure we do not have any regressions.
r=jgriffin
--HG--
extra : commitid : GkRPL9uVwjG
extra : rebase_source : 594ea1655f61552d430c3242e6b3418e550c27b3
is called. r=ato
This means that we are now doing step 7 in
http://w3c.github.io/webdriver/webdriver-spec.html#get
--HG--
extra : commitid : KQi5f9c0C1r
extra : rebase_source : 9c77a1d9aadf529ddbae2d32e0e1f9c4d532e181
extra : histedit_source : f40f377e1398f72fa68efd69ad864d9e4ae7e2b4