ElementArrayFinder.prototype.count is slow since 2.0.0

[Droogans]

Since 2a765c7 was included, I've seen a test that used to take 2 seconds now take almost 11.

This is the test.

it('should have 182 rows in the table', function () {
    expect(element.all(by.repeater('r in rows')).count()).to.eventually.equal(182);
});

The output from the Selenium server in version 2.1.0.

15:54:13.774 INFO - Executing: [is enabled: 0 [org.openqa.selenium.remote.RemoteWebElement@8dadb1bd -> unknown locator]])
15:54:13.782 INFO - Done: [is enabled: 0 [org.openqa.selenium.remote.RemoteWebElement@8dadb1bd -> unknown locator]]
15:54:13.786 INFO - Executing: [is enabled: 1 [org.openqa.selenium.remote.RemoteWebElement@f05f4f94 -> unknown locator]])
15:54:13.854 INFO - Done: [is enabled: 1 [org.openqa.selenium.remote.RemoteWebElement@f05f4f94 -> unknown locator]]
15:54:13.857 INFO - Executing: [is enabled: 2 [org.openqa.selenium.remote.RemoteWebElement@f2f33543 -> unknown locator]])
15:54:13.912 INFO - Done: [is enabled: 2 [org.openqa.selenium.remote.RemoteWebElement@f2f33543 -> unknown locator]]
15:54:13.915 INFO - Executing: [is enabled: 3 [org.openqa.selenium.remote.RemoteWebElement@cfff877 -> unknown locator]])
15:54:13.965 INFO - Done: [is enabled: 3 [org.openqa.selenium.remote.RemoteWebElement@cfff877 -> unknown locator]]
15:54:13.968 INFO - Executing: [is enabled: 4 [org.openqa.selenium.remote.RemoteWebElement@f8780d1a -> unknown locator]])
15:54:14.027 INFO - Done: [is enabled: 4 [org.openqa.selenium.remote.RemoteWebElement@f8780d1a -> unknown locator]]
15:54:14.030 INFO - Executing: [is enabled: 5 [org.openqa.selenium.remote.RemoteWebElement@e87b795b -> unknown locator]])
15:54:14.077 INFO - Done: [is enabled: 5 [org.openqa.selenium.remote.RemoteWebElement@e87b795b -> unknown locator]]
15:54:14.081 INFO - Executing: [is enabled: 6 [org.openqa.selenium.remote.RemoteWebElement@a47d2d50 -> unknown locator]])
15:54:14.127 INFO - Done: [is enabled: 6 [org.openqa.selenium.remote.RemoteWebElement@a47d2d50 -> unknown locator]]
15:54:14.130 INFO - Executing: [is enabled: 7 [org.openqa.selenium.remote.RemoteWebElement@ff6d7067 -> unknown locator]])
15:54:14.179 INFO - Done: [is enabled: 7 [org.openqa.selenium.remote.RemoteWebElement@ff6d7067 -> unknown locator]]

This continues until it reaches the end of the repeater.

Here's my webdriver-manager output on startup.

selenium standalone is up to date.
chromedriver is up to date.
seleniumProcess.pid: 55988
16:25:53.808 INFO - Launching a standalone server
Setting system property webdriver.chrome.driver to /Users/droogans/code/js/encore-ui/node_modules/protractor/selenium/chromedriver
16:25:53.851 INFO - Java: Oracle Corporation 25.45-b02
16:25:53.852 INFO - OS: Mac OS X 10.10.4 x86_64
16:25:53.863 INFO - v2.45.0, with Core v2.45.0. Built from revision 5017cb8
16:25:53.949 INFO - Default driver org.openqa.selenium.ie.InternetExplorerDriver registration is skipped: registration capabilities Capabilities [{ensureCleanSession=true, browserName=internet explorer, version=, platform=WINDOWS}] does not match with current platform: MAC
16:25:53.982 INFO - RemoteWebDriver instances should connect to: http://127.0.0.1:4444/wd/hub
16:25:53.983 INFO - Version Jetty/5.1.x
16:25:53.984 INFO - Started HttpContext[/selenium-server,/selenium-server]
16:25:54.018 INFO - Started org.openqa.jetty.jetty.servlet.ServletHandler@d7b1517
16:25:54.018 INFO - Started HttpContext[/wd,/wd]
16:25:54.018 INFO - Started HttpContext[/selenium-server/driver,/selenium-server/driver]
16:25:54.018 INFO - Started HttpContext[/,/]
16:25:54.023 INFO - Started SocketListener on 0.0.0.0:4444
16:25:54.024 INFO - Started org.openqa.jetty.jetty.Server@7cd84586

I'm using firefox version 35 to run these tests. Using ChromeDriver 2.15.322455 (ae8db840dac8d0c453355d3d922c91adfb61df8f) outputs the same logs, but runs much faster at just over 3 seconds.

[sjelin]

Thanks for the report, but it's not clear if we can do anything. This sounds like an issue with FireFox driver, which would be out of our control.

[Droogans]

I made a comment on 2a765c7#commitcomment-12254207 for an idea of what I think the problem might be.

Why check if the list has stale elements when counting? Why not expect the user to ask for a new ElementArrayFinder every time? The way the unit test was written implies that protractor will double check your side effects for you.

[hankduan]

Thanks @Droogans. This should help: #2364

[hankduan]

fixed with #2364