Wednesday, October 14, 2009

A Small Research on Selenium



I had a great curiosity to analyze how the time elapsed for a test varies with the speed. So I conclude a small research

Environment:

Notebook Asus X51RL

WSO2 Governance Registry 3.0.1

Test suite = Notification test

https://wso2.org/repos/wso2/trunk/commons/qa/web-test-framework/registry/src/test/java/org/wso2/carbon/web/test/registry/NotificationTest.java

This test does not require an Internet connection. So the elapsed time was totally depending on the performance of the hardware and governance server only. Every time I used a fresh instance. This test has 30 test cases and each time test errors and test failures were recorded.

No test failures and no test errors were observed.

The speed varying graph has the shape of y=mx+c. C is depending on the hardware and Greg server and even selenium server. M is roughly 144 s/speed unit in hundreds. I couldn’t test m is hardware independent or hardware dependent.

Programmers frequently use Thread.sleep() to eliminate errors and failures. So there is a constant delay . I measured it. I replace the Thread.sleep() with the Thread.sleep();System.out.println(); I found there is a constant over head of 417 s.


No comments:

Post a Comment