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
http://example.com for testing and we
find in the sitemap the two URLs
http://foobar.com/bar/. 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.
http://econom.ía.com. The character following
econom. is not the lowercase English vowel
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.xn--a-iga.com. In the latter, the character
that looks like an
i but which is not is encoded as
In fact, if you submit
http://econom.ía.com 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.ía.com/example match the domain name
http://econom.xn--a-iga.com? If you compare them literally,
the answer is no. That's what we were doing.
If you were to submit
http://econom.xn--a-iga.com for testing
and the sitemap contained a URL of
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.