Released Selenide 5.18.0

Released Selenide 5.21.0

Blog

Released Selenide 5.18.0

Say no to webdriver logs

Good evening!

Today is January, 23.

Let’s meet the new update: Selenide 5.18.0.

Let me invite you to take a small virtual tour of the changes in 5.18.0.
Make yourself comfortable.


We disabled webdriver logs by default

Starting from version 5.13.0, Selenide wrote webdriver logs to files build/reports/tests/webdriver.uuid.log.
It seemed to be useful at that moment. But later we realized that these logs take quite a lot of disk space, and nobody really needs to read them in most cases. That’s why we decided to disable them by default.

If you need those webdriver logs, you can still enable them with setting Configuration.webdriverLogsEnabled = true.

See issue 1365 and PR 1379.


Changed “timeout” parameter type from Long to Duration

… in collection methods. Now instead of

  $$.shouldHave(texts(...), 5000);

it is fashionable to write

  $$.shouldHave(texts(...), Duration.ofSeconds(5));

See issue 1377.
Thanks to Ostap Oleksyn for PR 1377.


Speed up search of inner shadow dom elements

One of features that do exist in Selenide, but not in Selenium webdriver is shadow dom:

$(shadowCss("p", "#shadow-host", "#inner-shadow-host"))
    .shouldHave(text("The Shadow-DOM inside another shadow tree"));

There might be multiple shadow roots inside of other shadow roots.
To find all elements inside inner shadow roots, method $(shadowCss()) performed a JavaScript snippet in loop for every shadow root separately. It might be slow because every call to webdriver takes time.

Now $(shadowCss()) calls one even more trickier recursive JavaScript which finds all elements in all inner shadow roots with one call.

See PR 1373.

A separate “thank you” goes to sakamoto66 for issue 1246 and PR 1233 which unfortunately was not merged, but inspired us for this optimization.


Fixed checks $.shouldNot(and(...)) and $.shouldNot(or(...))

Selenide user pavelpp detected a bug in Selenide when used not in combination with and and or:

$(".lolkek").shouldNotBe(visible, ofSeconds(5));   // works
$(".lolkek").shouldNotBe(and("foo", visible));     // fails
$(".lolkek").shouldNotBe(or("foo", visible));      // fails

Now we fixed it. All three lines don’t throw errors anymore.

See issue 1369 and PR 1370.

By the way, we now forbid using and and or with only one condition. In other words, line or("foo", visible) doesn’t compile anymore. You have to use at least two condition, like or("foo", visible, enabled).

Agree, this is logical.


We detect conflict in “browserName” capability

We recently realized that there are two settings for specifying browser:

  1. The primary - Configuration.browser
  2. Another - Configuration.browserCapabilities["browserName"] (I don’t really know why it’s needed)

And you might get an unexpected browser opened if you haven’t set the first one, but set the second one. Now Selenide detects such a conflict and throws an exception:

IllegalArgumentException: Conflicting browser name: 'chrome' vs. 'firefox'

Let me repeat: setting Configuration.browser is enough in all cases. As far as I know, the second setting is never needed.

See issue 1366 and PR 1374.


Fixed displaying timeout parameter in reports

Recently we added to $.should* methods parameter timeout of type Duration.
Then we realized it was displayed quite unclearly in Selenide/Allure reports.

For example, line $("h1").shouldBe(visible, Duration.ofSeconds(1)) looked like this in a report:

$("h1") should be([visible, PT1M])

Though this “PT1M” is an ISO-compatible representation and means “time period 1 minute”, we decided to replace it by more familiar “300 ms”, “1s”, “1.500 s.” etc.

See issue 1376 and PR 1378.


Upgraded to WebDriverManager 4.3.1

As usually, see their changelog here.


It’s the end of our excursion. Feel free to update and share your feedback.
Create issues in github, complain in chats, swear in twitter.

News

Here is an final downloads statistics of Selenide for year 2020. Not bad!


Happy current year!

Andrei Solntsev

selenide.org

blog comments powered by Disqus

The project is maintained by