SELENIUM 3 THINGS YOU NEED TO KNOW

Posted by Infocampus HR on September 6th, 2018

SELENIUM 3 THINGS YOU NEED TO KNOW

Selenium can be a standout amongst the most great instruments in the QA Engineers' munititions stockpile. Lamentably, it tends to be tedious to compose, a bad dream to keep up, and it can deliver far an excessive number of false negatives. At Lucid Software, that is the place our suite was amidst 2014.

We had a suite of around 300 tests composed by 40 distinct designers averaging 60-70 disappointments every night. Accountable for it, we had 1 committed full time design whose activity was just to deal with which tests were honest to goodness disappointments and which tests were false negatives. This would habitually take his whole work day. At the point when tests were ended up being excessively problematic, they were boycotted and pushed off into the heap of different errands to be finished "sometime in the future."

My enthusiasm for Selenium was provoked when an element that I had included, alongside the test for the component, broke. We didn't discover the bug until the point when it had just been underway for about fourteen days. There was positively no reason for a bug like this to ever influence it past improvement, to let alone through QA and into generation. I took a gander at what had happened to the test: it turned out it had been boycotted (placed in a rundown of skipped tests) about multi month after I composed it. The inquiry emerged: what should be done to influence our Selenium to test suite solid, adaptable, and viable?

Make Tests Easier to Write

The number 1 objection we had from designers around composing tests was that it accepting the same amount of work to compose the tests as it did to settle the bug or include the component. In light of that, we realized that we needed to make tests simpler to compose. This prompted the production of substances like  Seleniun Training In Marathahalli Application User and Application Driver.

1. Make An Application User

An Application User is the Selenium portrayal of the backend of the site. For us, this implied making another Lucidchart or Lucidpress client, setting up a membership, and potentially making a report. This class contained the partner techniques to set up a test situation and all the teardown toward the finish of a test. This class additionally contained access to our backend administrations to make things, for example, including colleagues, transferring pictures or textual styles, and changing membership levels extremely simple. The accompanying is a case of how an engineer would utilize the application client.

In this circumstance, all setup was disentangled to two simple technique calls, leaving the test prepared to go in the manager. Toward the finish of the test, the majority of the tear down (shutting the driver, database expulsion, and so on) is dealt with in the completed technique. By abstracting the greater part of this, (and in addition a few other partner techniques) into a User class, we made it significantly less demanding for designers to get a test set up and prepared to approve a bug or highlight.

2. Make an Application Driver

The Selenium API can be exceptionally overwhelming. There are around 20 distinctive approaches to get a component utilizing the WebDriver. From that point, there are incalculable approaches to perform diverse activities, for example, relocating, clicking and right clicking, utilizing the parchment wheel, and composing. With an end goal to disentangle this, and make it so every one of the engineers require not get comfortable with the whole WebDriver documentation, we made a driver to streamline the most widely recognized activities. This application driver expands WebDriver and includes the Selenium activities class. From that point, we included techniques joining the most widely recognized activities, for example, tapping on a component, executing a content, and moving web components. The class looked something like the accompanying UML outline, and contained exceptionally basic strategies, for example, the ones demonstrated as follows.

At the point when engineers expected to accomplish more intricate activities, despite everything they approached the WebDriver and Actions classes, yet for most tests, the constrained usefulness of the Lucid Driver was all that anyone could need. This had the special reward of making tests significantly simpler to investigate in light of the fact that the greater part of the engineers were presently utilizing similar strategies rather than every designer looking through the Selenium API and finding an alternate method to perform indistinguishable usefulness.

Make Tests Easier to Update

With a consistently changing item and numerous designers including and refreshing highlights, it is simple for tests to wind up obsolete. At the point when a component was refreshed, we required an approach to rapidly port the old tests to work with the refreshed DOM. Utilizing DOM IDs and the page protest design helped us make tests that were effortlessly refreshed and viable.

3. Utilize DOM IDs

Finding a component in the DOM can be a standout amongst the most difficult parts of a Selenium test. IDs give an approach to key components to be extraordinarily distinguished inside the whole item. In a portion of our unique tests, we utilized XPaths, class ways, and other complex CSS selectors to find vital components. Nonetheless, when a component moved to a better place in the UI, or essentially simply changed its CSS class names (due to overhaul or refactoring), refreshing the test required returning and finding that component once more. With IDs, a component is identifiable paying little mind to where it is in the DOM and what styling is connected to it.

The following is a case of a noteworthy port we did in the Publish discourse of Lucidpress. This specific component had 4 Selenium test suites, or around 30 tests altogether. There were furthermore another 20-30 tests that utilized the discourse somehow. Since we utilized IDs, the vast majority of the tests required practically zero refreshing. The test could discover the distribute catch, create code catch, estimate selectors, and other key components with no issues. This turned what could have been for all intents and purposes revising every one of the tests, into essentially rolling out a few improvements in how to explore to the right tab of the discourse.