372401 - Thunderbird/Firefox/Gecko's spell checking engine should check DNS if it thinks the name is a domain

Status: NEW

(Core :: Spelling checker, enhancement)

Description

[Gregory Gerard]

User-Agent:       Opera/9.10 (Windows NT 5.1; U; en)
Build Identifier: beta 2

www.burningsea.com gives me the red underline of grammatical naughtiness.

I'd think that if it looks like a web address (DNS legal words separated only by '.') Thunderbird could either just ignore this or do something cooler which would be to auto extend it's dictionary with whatever comes back in a DNS request.

Reproducible: Always

Steps to Reproduce:
Type in a domain name that is not in the dictionary.
Actual Results:  
You get the red underline of grammatical naughtiness.

Expected Results:  
Since DNS is actually a ginormous distributed dictionary (albeit only not always available), the Gecko spell checking engine should use this automatically (or maybe through a preference defaulted to do so automatically). Why distract the user with stuff that is actually verifiably, demonstrably, and cheaply correct? False negatives in a spell checker are a bane. Let's take Outlook/Word for instance...

Comment 1

[Brett Wilson]

There is no way it can check DNS for everything that might be a domain. Performance would be awful and it would tie up your network. There is also no way there will be a preference added for this. These requests would also have to be asynchronous which is not the way the spellchecker is designed.

If you don't like these, you can type "http://" in front and it will be identified as a URL.

It would be feasible to hook it up to the tld service so that if you have a sequence of words separated by dots, and the ending is a known tld, we can assume its a URL.

Comment 2

[Andy Behrens]

Shouldn't this be listed as an enhancement rather than a bug?

At ant rate, I'd rather see this as an application plugin rather than part of gecko's core. It's just too far removed from gecko's primary function of rendering html.

Comment 3

[Matt Caywood]

Even better would be an extension that ensured that URLs were both well-formed and linked to valid sites!

Perhaps simpler, the spellchecker could just accept www(.*)* as correct.