Simply Testable Blog

Figuring out how to automate away the pain of routine front-end web testing; the story behind

211 posts covering the initial idea, growth of the service, features, advances, failures and successes.

Thanks to the feedback from @igll, internationalised domain name (IDN) support has been improved.

I'll explain now what the problem was and how it has been fixed.

When discovering the URLs to test for a given site, the URLs that are discovered are compared against the site that was submitted for testing so that we only try to test URLs that are relevant.

Let's say you submit for testing and we find in the sitemap the two URLs and We'll compare the domain of the site submitted for testing with the domain name in the URLs we find and discard those that don't match. In this case, the second URL would be discarded.

We weren't correctly taking into consideration the two ways you can represent an IDN.

Consider http://econom.í The character following econom. is not the lowercase English vowel i, it's a character that looks similar but which has an accent instead of a dot.

í vs i. See? Good, back to the point.

You can present the above URL as http://econom.í or In the latter, the character that looks like an i but which is not is encoded as xn--a-i.

In fact, if you submit http://econom.í for testing, our system see this as you submitting for testing as, behind the scenes, this is what computers really see.

When comparing discovered URLs with the site that was submitted for testing, we weren't comparing things quite right. In some cases, we were asking: Does http://econom.í match the domain name If you compare them literally, the answer is no. That's what we were doing.

If you were to submit for testing and the sitemap contained a URL of http://econom.í, we would previously have discarded it as the domain names didn't match.

We now compare the domain names in URLs in an IDN-aware manner so that such URLs are not discarded.