[whatwg] Spellchecking mark III

Peter Kasting pkasting at google.com
Wed Jan 28 09:47:00 PST 2009

On Wed, Jan 28, 2009 at 2:35 AM, Křištof Želechovski <kkz at mimuw.edu.pl>wrote:

>  *No, the _original_ use was to turn it on on fields where it would
> otherwise have been on.
> *
> I do not understand.  If spell checking would be on, why turn it on
> explicitly?
I mistyped.  The last word should have been "off".

> If the control is not expected to contain a private language, it should be
> subject to spell checking.
This thread has already had multiple examples of cases where this is untrue.
 Spelling quizzes, address fields, etc.  And even if it were true, it's not
the way browsers behave today (e.g. Firefox does not spellcheck single-line
fields, precisely to avoid a lot of cases like this), and changing those
defaults to be something non-annoying, using complex heuristics, is
significantly harder (in terms of your time/money cost below) compared to
supporting the attribute.

>     Avoiding an additional attribute is a gain,
>  Why?
> Because adding an additional attribute costs time and money.
To whom?  What tradeoff are you making?  Keying spellchecking off language
support costs engineering time too, for the UA.  And for a web author.  All
changes have costs.  The point here seems like a vague principle rather than
a specific application.

>   Which no one will ever use, because users aren't going to take the
> trouble to declare such a thing when human recipients can just _read the
> text_.  After all, WE have built-in language detectors in our heads.
> We disagree here but further discussion is void unless you have the
> resources necessary to perform an investigation of the subject.
If you need data to prove that people will not make the effort to explicitly
tell recipients what languages their messages are in, I offer you the entire
history of written communication, where people don't say "By the way this is
in English!" at the top of each letter.

> Users entering text in a foreign language cause trouble to the forum
> moderators who have to discipline them.  Thus, the software could
> accommodate to the needs of the moderator, so that the poster gets warned
> before posting, not admonished afterwards.  This is more convenient and less
> work for everyone.  Providing an indication what language is recommended by
> forum users is good, because most users would take that into account (for
> fear of getting plonked, if not for good manners).
How is this relevant to a discussion about spellchecking?  If you want
UA-based language detection facilities that are, say, accessible from JS,
that may be a reasonable request, but like much of this discussion, it seems

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/attachments/20090128/9aea7224/attachment-0001.htm>

More information about the whatwg mailing list