
It's better to define a custom name for each, for easier processing. If you start 3 runs with 3 different browsers, and don't bother copying the files, they will be overwritten. Thus you can start one run with Chrome, another with Firefox, and so on.Ī small tip - by default, the run logs are in files with the names "output.xml", "log.html" and "report.html".

This btw is described in details in the user guide, the Variables section. The argument -variable is used to set the value of one it is given in the format var_name:value. If it's not overridden, it has a default value (Chrome in that case).Īnd now, to run with a different browser you just set it its name, on the CLI for starting a run sample: robot -variable browser:Firefox suites\sample.robot Suite Setup Open Browser url= browser=$ variable.

# the browser will be opened in the start-up of the suite
Multibrowser testing how to#
Thus a full coverage matrix will be produced - the run on one browser doesn't affect the run on another.Ī very basic sample how to do it - code with inline comments: *** Settings *** The usual approach for multi-browser testing is to run the collection of cases, one by one, against each browser.

The feature does sound quite cool, but will require quite some modifications in it (all revolving around relaying the same command to all browsers), and taking care of any unforeseen circumstances in its current architecture (what should happen if one of the selenium drivers loses connectivity, but the others are working ok? etc). The SeleniumLibrary doesn't have the capability to run the same command against multiple browsers.
