Unless you have been under a sizeable rock for the last 2+ years, you have to be aware of the phenomenon that is Stack Overflow. One offshoot of its success has been branded versions called Stack Exchanges. The Selenium project has a proposal for one that has been languishing for some while and I suspect its inability to get over the hump is what prompted this to be put at the top of the proposal today.

This could be the kiss of death for the idea. And is a

Very.

Bad. (Dumb? Ill informed? No… let’s stick with Bad.)

Idea.

And illustrates just how poorly we have explained what Selenium is, is not, and where it is going. Let’s start with the not one first since it is the source of all this confusion.

Selenium is NOT a testing tool.

Huh? But I use it for testing! you might be saying. Without getting into the Testing vs. Checking debate, I’ll say what Selenium is which might help clarify things.

Selenium is a BROWSER AUTOMATION tool.

(Well, tools, libraries and APIs, but this is close enough.)

Yes, you might use Selenium to aid in your Testing activities but it is just its use in your context that makes it a testing tool. I know of large Drupal sites thats that use it for administration tasks across many instances; does that make it an DevOps tool rather than a Testing one? No. How about downstream projects that incorporate parts of Selenium as a building block? Does that make Selenium a core library like GLibC? No. The tool remains the same, but its use is context specific.

Merging the Selenium proposal in with ‘Software Testing and QA’ one completely erases the contextual importance of its use cases.

If anything, I think the current Selenium proposal should be widened to include that Watir community since we have started the reunification around WebDriver already.

Where Selenium is going is another reason it having a distinct Stack Exchange from ‘Software Testing and QA’. There are big plans afoot for the use of the WebDriver API to be the[!] standard for automating a browser. Opera has already released a Watir driver for its flagship browser (and Watir is driven by WebDriver these days), Chromium (the Open Source version of Google Chrome) has WebDriver hooks in trunk and we’re working with Mozilla to get WebDriver integrated into their Mozmill framework. Safari will likely grow this functionality through WebKit and Internet Explorer, well, once Microsoft decides to really participate in Selenium, they’ll likely see the value in this and hop on board as well.

So does it make sense for people doing the integrations like this to go to a Testing Stack Exchange? Of course not. These people won’t likely self-identify as testers and think to go there.

Nor will the users of the other contexts either. Any why would they? They are not testing with it; they are automating a browser. Which means that you splinter the information across multiple Stack Exchanges instead of just a single one.

Which would be a worse outcome in the long run.

In essence, this was a long winded way of saying:

  • Merging the Selenium Stack Exchange proposal with a generic ‘Software Testing and QA’ one completely misses the point about what Selenium is now, and will be in the future
  • It should be expanded in scope to include Watir
  • If it does get merged with something else, Selenium users (of all contexts) should be redirected to Stack Overflow and not another Stack Exchange to keep things in a single ‘stack’.