[UA-discuss] [UA-EAI] Request for review: Report on Evaluation of Websites for Acceptance of E-mail Addresses, 2019

[Marc Blanchet]

On 12 Jul 2019, at 16:20, John R. Levine wrote:

> On Fri, 12 Jul 2019, Marc Blanchet wrote:
>>> With respect to the HTML5 pattern for e-mail addresses, I have 
>>> talked to people in WHATWG about it.  That pattern is correct for 
>>> ASCII addresses and it's not going to change to an EAI pattern 
>>> because that would lead to web sites with ASCII mail systems 
>>> accepting addresses to which they can't send mail.  They would be 
>>> open to adding a new "eaimail" input type that accepts EAI 
>>> addresses, to allow an easy upgrade when sites have EAI capable back 
>>> ends.
>>>
>>> Nothing is going to happen in WHATWG until one of their large 
>>> members says they'll implement it which hasn't happened.  I've made 
>>> some inquiries and gotten polite responses, but I can't do much more 
>>> since I have no funding.
>>
>> from 
>> https://w3c.github.io/test-results/html53/implementation-report.html, 
>> done back in september, Firefox seems to support, while others were 
>> not tested.
>
> Ah, that is as we say a can of worms.

I know. actually, the table states that Firefox works, but my testing 
shows otherwise on my environment. And the table does not tell version 
number, platform, etc…

>
> The actual HTML spec that major browser vendors implement is the 
> WHATWG living standard.  W3C copies that spec verbatim into their own 
> standard, except that they make some incompatible changes.  WHATWG has 
> repeatedly asked W3C not to do that, but W3C persists.

yeah. I know all that story and we talked about it extensively when I 
was on the IAB…

Marc.

>
> One of those incompatible changes is e-mail addresses.  WHATWG has a 
> recommended address validation pattern pattern which is a subset of 
> the RFC5321/5322 spec that matches what real mail systems accept well. 
>  W3C changed that to accept any UTF-8 which is wrong for many reasons. 
>  One is that it's gratuitously incompatible, another is that it 
> provides no way to distinguish between EAI and non-EAI back end mail 
> systems, and a third is that real EAI mail systems are unlikely to 
> accept all of the random mixes of scripts and punctuation that the W3C 
> pattern allows.
>
> The way people use the WHATWG pattern is that they cut and paste it 
> into their javascript libraries and the browser just runs the 
> javascript.  So long as everyone follows WHATWG it hardly matters what 
> W3C does.  Before WHATWG will add an EAI pattern for programmers to 
> use, it needs an eaimail input type to distinguish EAI and non-EAI 
> mail backends, which needs browser and web server support.
>
> Regards,
> John Levine, johnl at iecc.com, Primary Perpetrator of "The Internet for 
> Dummies",
> Please consider the environment before reading this e-mail. 
> https://jl.ly