Fix ChromeVox next tests that passed without actually running.

Fix 2 ChromeVox next tests that passed without actually running.

The way to create callbacks for async functions in these e2e tests is to
invoke this.newCallback().  See https://codereview.chromium.org/938623003
for more background.  If this is not done, the tests will pass silently
before the asyc callbacks get called which means much of the tests don't
run.

This CL fixes two of the tests and marks the rest with TODOs because
they fail and need more debugging when this problem is fixed.

The live region tests are fixed by making paragraph nodes output their descendants
instead of their value property.

BUG=None

Committed: https://crrev.com/3d95fe156c4ebc4bec9c476a02761a102a7316b8
Cr-Commit-Position: refs/heads/master@{#343217}

[Peter Lundblad, 2015-08-13 15:56:51]

plundblad@chromium.org changed reviewers:
+ dmazzoni@chromium.org, dtseng@chromium.org

[Peter Lundblad, 2015-08-13 15:56:52]

I haven't figured out yet if the other tests that fail are caused by actual cvox
next bugs or races in the test.

[dmazzoni, 2015-08-13 16:04:24]

The CQ bit was checked by dmazzoni@chromium.org

[dmazzoni, 2015-08-13 16:04:25]

lgtm

Thank you!

Good timing, I started writing a test this morning and discovered the same
thing.

Any ideas for how to make this more foolproof so it's harder to write (or
modify) a test in a way that cause it to succeed without actually doing
anything?

[commit-bot: I haz the power, 2015-08-13 16:04:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1288043002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1288043002/20001

[commit-bot: I haz the power, 2015-08-13 16:35:49]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-08-13 16:36:26]

Patchset 2 (id:??) landed as
https://crrev.com/3d95fe156c4ebc4bec9c476a02761a102a7316b8
Cr-Commit-Position: refs/heads/master@{#343217}

[Peter Lundblad, 2015-08-14 13:37:23]

dmazzoni@chromium.org writes:
> Any ideas for how to make this more foolproof so it's harder to write (or
> modify) a test in a way that cause it to succeed without actually doing
> anything?

In general, I think that's hard.  What I usually do when writing tests,
especially async ones, is to introduce something that causes them to fail
(often a typo in the expected output) and make sure they fail.
Of course that's manual and error-prone.

For the particular case, I killed testDone in our tests totally.  See
https://codereview.chromium.org/1295773002.
Hopefully it will prevent us from falling back into old habbits when writing
new tests.

Thanks,
//peter

> https://codereview.chromium.org/1288043002/

-- 

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.