Re: Re: Guessing the fallback encoding from the top-level domain name before trying to guess from the browser localization

On Thu, Jan 2, 2014 at 10:02 PM, Liam R E Quin <liam@w3.org> wrote:
> Note that there are plenty of other cases where an encoding is wrongly
> marked that this change would not solve.

True. As seen in Firefox telemetry, the most common case is overriding
the encoding of a labeled page. But it's unclear how much that really
helps, considering that the next most common case is overriding a
previous override (i.e. the previous attempt at overriding evidently
was not successful).

> Solving the most common case is
> a good thing to do, but it won't obsolete the need for an override menu.

Outside developer tools that allow arbitrary DOM editing browsers
don't provide user override UI except for two classes of authoring
error: bogus character encodings and bogus TLS certificates. And for
both of these, there is already a way for the author to remove the
override UI (BOM and HSTS, respectively). There's a vast variety of
other kinds of authoring errors for which everyone seems to accept
that browsers don't provide UI for users to repair severe authoring
errors. That is, not providing UI for fixing authoring errors is the
norm.

At some point, mis-labeling the encoding might be able to join that
category of errors. However, *politically* for that to happen, the
frequency of encoding menu usage in the CJK locales needs to become
similar to other locales. I expect this feature to help on that point.
Once the notion that removing the menu is would disproportionately
hurt particular locales goes away, we can wait until we see enough
nines after 99.99% unuse and eventually decide the menu is unused
enough not to be worthwhile anymore. (Chrome already buries it so deep
in the submenus that one has to wonder if it's worthwhile to have UI
that's so hidden away from users.)

-- 
Henri Sivonen
hsivonen@hsivonen.fi
https://hsivonen.fi/

Received on Friday, 3 January 2014 14:45:00 UTC