Domain validator now too permissive

[leforestier]

Hello,

It seems any string composed of numbers separated by dots is now accepted by the domain validator.

>>> validators.domain("123.123")
True

>>> validators.domain("123.123.123")
True

>>> validators.domain("123.123.123.123")
True

It didn't use to be the case (I know because it now causes some errors in my unit tests in the "naval" library)
I think it was good that these strings weren't accepted as domains. I don't think these are valid domains.

[alertedsnake]

Confirmed this is a regression in 0.13.0 - it works in 0.12.5.

[alertedsnake]

Although... is it a regression? Seems like a philosophical question: "Is an IP address a valid domain name?" Hmm...

[micahcochran]

Although... is it a regression? Seems like a philosophical question: "Is an IP address a valid domain name?" Hmm...

.123 is not a valid Top Level Domain. Furthermore, based on RFC 1034's rules it looks like a valid domain name (page 11) have to begin with a letter.

Edit: Correct page number and add anchor hyperlink.

[alertedsnake]

In digging further I believe this address is not a valid domain name due to the TLD not being valid as @micahcochran said, however the matter of a valid domain name starting with a number is less clear - RFC2181 suggests othewise, and in practice, these exist in the wild.