Looks like a slow test (on Chrome dev, at least):

 1:07.56 INFO ## Slow tests ##

 1:07.56 INFO |                     Test                    | Result | Longest duration (ms) | Timeout (ms) |
 1:07.56 INFO |---------------------------------------------|--------|-----------------------|--------------|
 1:07.56 INFO | `/media-capabilities/decodingInfo.any.html` | `OK`   | `10109`               | `10000`      |
 1:07.56 INFO 
 1:07.56 INFO ::: Running tests in a loop 10 times : PASS
 1:07.56 INFO ::: Running tests in a loop with restarts 5 times : FAIL
 1:07.56 INFO :::
 1:07.56 ERROR ::: Test verification FAIL

I suspect the test was already slow based on the changes in content, and we had just been scraping the edge for a while. Need to check if its marked slow already - if not, then let's do that. In either case, will warn the author of the CL after we mark the test as slow. @KyleJu do you have bandwidth available to do that?

Looks like a slow test (on Chrome dev, at least):

 1:07.56 INFO ## Slow tests ##

 1:07.56 INFO |                     Test                    | Result | Longest duration (ms) | Timeout (ms) |
 1:07.56 INFO |---------------------------------------------|--------|-----------------------|--------------|
 1:07.56 INFO | `/media-capabilities/decodingInfo.any.html` | `OK`   | `10109`               | `10000`      |
 1:07.56 INFO 
 1:07.56 INFO ::: Running tests in a loop 10 times : PASS
 1:07.56 INFO ::: Running tests in a loop with restarts 5 times : FAIL
 1:07.56 INFO :::
 1:07.56 ERROR ::: Test verification FAIL

I suspect the test was already slow based on the changes in content, and we had just been scraping the edge for a while. Need to check if its marked slow already - if not, then let's do that. In either case, will warn the author of the CL after we mark the test as slow. @KyleJu do you have bandwidth available to do that?

Marked as a slow test and left a comment on the CL

@KyleJu @stephenmcgruer how does "Running tests in a loop with restarts 5 times" translates in Chrome's web_tests runner? I can't reproduce this and I wonder if I don't use the right configs or if it's because WPT is using //chrome while we do not internally in Blink. @Rob--W @foolip FYI

@chcunningham I wonder if the issue could be the loading of the DB assuming the above is correct and it can't be repro in Blink.

Hey, sorry for the delay @mounirlamouri . "Running tests in a loop with restarts 5 times" would be approximately equivalent to passing --repeat-each=5 and --restart-shell-between-tests to run_web_tests.py (cc @LukeZielinski). However what we're actually looking for here is that any single run of the test took long enough that it threatened to timeout, so a single run should reproduce it.

Locally running this a few times, I got one result of 9.8 seconds which is close to the timeout, and one result of 10.6 that is over it:

...
Ran 1 tests finished in 9.8 seconds.
$ ./wpt run --no-pause-after-test --binary ~/chromium/src/out/Release/chrome chrome media-capabilities/decodingInfo.any.html
...
Ran 1 tests finished in 10.6 seconds.

Now I'm using a local build so not as fast as a release one (which is what wpt.fyi uses), but still a quite slow test!

Also if you look at https://wpt.fyi/results/media-capabilities/decodingInfo.any.html?label=experimental&label=master&aligned, Chrome took 2.927 seconds to run this test file (on the run I looked at) vs 0.42 seconds for Firefox.