Open Bug 1731034 Opened 1 year ago Updated 1 year ago

Consider extending domainsuffixwhitelist with common invalid / RFC2606b-proposed

Categories

(Firefox :: Address Bar, enhancement, P5)

enhancement

Tracking

()

Tracking Status
firefox94 --- affected

People

(Reporter: gcp, Unassigned)

References

Details

.lan is a common suffix is used by routers, users seem to be running into this as can be observed by Google searches and SUMO.

Some digging finds this proposed (but I guess, not accepted? RFC):
https://tools.ietf.org/id/draft-chapin-rfc2606bis-00.html

Which links to:
https://www.icann.org/en/system/files/files/sac-045-en.pdf

Also:
https://datatracker.ietf.org/doc/html/rfc6762#appendix-G

Based on this, could we consider extending the default whitelist at
https://searchfox.org/mozilla-central/rev/4d90ff4537330d6b17cb956c0fadf759086d9bb7/browser/app/profile/firefox.js#245
with:

.intranet
.private
.corp
.home
.lan
.localdomain
.domain
.host
.invalid
.wpad
.belkin

For reasons explained above, they're unlikely to become real TLD, and given that this is a pref, we can easily adjust if needed.

See Also: → 1634650

We are already supporting .test, .example, .invalid, .localhost, .internal and .local that are pretty much covering most needs.
There must be some kind of rule, otherwise we risk that any kind of proposal will force us to add even more suffixes. So far that rule has been that we'll pick RFC defined suffixes.

I'm not sure we want to go much further with the hardcoded list (unless this RFC proposal one day gets approved, but it looks like it elapsed), especially with it containing suffixes named after companies. We could maybe accept some names, if there would be updated statistics showing they are largely in use. Without clear evidence we could just make good or bad guesses.

I think that, at a certain point, we should instead improve our "did you mean to go to" notification bar to happen not just when I type a domain but also when I type a path after it (I think there's a bug on file, but maybe I'm misremembering cause I can't find it), and implement Bug 1635062, so it's easier to add custom lists of suffixes through an add-on.

Severity: -- → N/A
Priority: -- → P5

We could maybe accept some names, if there would be updated statistics showing they are largely in use. Without clear evidence we could just make good or bad guesses.

See the SAC045 document from ICANN, which provides exactly that data.

It is data from 12 years ago though (collected until 2009). It is bringing up a very valid point of course, but its recommendation was not fullfilled, and I don't know why.
We already support the top 2 from that list, local and localhost. We should keep monitoring the situation and see if new recommendations come out, otherwise we should telemetry Firefox most common invalid tlds and cross check with this list to do a good job.

we should telemetry Firefox most common invalid tlds

Are we able to collect this kind of data? I can see checking against that list (we could at least gather how much % of "wrong" queries match that list), but getting new unknown URLs from the user seems tricky?

You need to log in before you can comment on or make changes to this bug.