- From: Robert J Burns <rob@robburns.com>
- Date: Fri, 31 Oct 2008 13:34:03 -0500
- To: Philip Taylor <pjt47@cam.ac.uk>
- Cc: Chris Wilson <Chris.Wilson@microsoft.com>, HTML WG <public-html@w3.org>
- Message-Id: <4A0A8CF7-2525-4CF7-9F56-801B85B7753F@robburns.com>
HI Philip,
On Oct 31, 2008, at 12:08 PM, Philip Taylor wrote:
>
> Chris Wilson wrote:
>> Ben Millard [mailto:cerbera@projectcerbera.com] wrote:
>>> C. Generated quotes can be altered from CSS by at least Firefox
>>> and Opera (I
>>> tested Firefox 2.0.0.17 and Opera 9.60) but the default characters
>>> they
>>> generate are different:
>>> <http://meyerweb.com/eric/css/tests/css2/sec12-04-01.htm>
>>> <http://lists.w3.org/Archives/Public/public-html/2008Oct/
>>> 0103.html>
>> Yes. Generated quotes can be altered from CSS in IE8 as well. We
>> appear to match Firefox 3 behavior on Eric's tests, to be specific
>> - that is, we use non-ASCII quotes by default.
>>> D. Firefox makes a token effort to support one non-Enlish language
>>> in one
>>> localisation:
>>> <http://lists.w3.org/Archives/Public/public-html/2008Oct/
>>> 0102.html>
>> We do better than that. I'd have to look up precisely how broad it
>> is, but we recognize a set of lang values.
>
> (I can see 21 different language-dependent quote styles in IE8b2,
> which does seem much better than a token effort.)
I think the token effort accusation was against Firefox, but I think
that was also based on a misunderstanding which Boris later corrected.
>
>
> What is IE meant to do by default when no language is specified?
>
> If I write <!DOCTYPE html><q>Test</q> in IE8b2, then it outputs
> single-quotes. <q lang=en-gb> is the same. lang=en and en-us output
> double-quotes instead.
>
> I'm running on an en-GB system, so I guess IE must be picking up the
> system language and using that as the default. (I could be wrong (in
> which case this whole post is irrelevant) since I can't find an easy
> way to test this, but I don't see any other reason it would default
> to something equivalent to en-GB.)
>
> That seems bad because it will be unexpected and surprising to
> authors. A likely situation is:
>
> An en-US speaking author writes a perfectly valid page in English,
> with a doctype that triggers IE8-compatibility mode in IE8, without
> setting lang. They read somewhere that <q> puts quotation marks
> around quotations by default, so they use it for their page's
> quotations (or for citations or for emphasis or whatever). They test
> it in every browser they have access to, and it creates double-
> quotes in all of them, so they're happy and they don't need to
> fiddle with CSS and they publish the page. Then a foreign (French,
> Japanese, etc) user (who can read English perfectly well) visits the
> page, and their browser defaults to their system language and draws
> funny quotes in the middle of the English page, which makes the page
> look strange and broken, and everyone is unhappy.
>
> To avoid that problem, the quote style should be purely a function
> of the page, not of the user. When no lang is specified in the page
> (or in HTTP headers), the quotes should default to something
> consistent (probably en-US because that's what all other browsers
> do). Then authors will test their page, and non-en-US authors will
> see that it's not quite what they want and will fix it by specifying
> lang="..." or "q { quotes: ... }" or by using inline punctuation
> instead of <q>, and then it will work exactly the same for all users.
>
> This increases the chance that the default will not match what the
> author wants, so they will have to expend some minor effort to fix
> it; but it avoids the much larger pain of them thinking their page
> looks how they want it to, when actually many users will see
> something quite different.
>
> Does something like this seem reasonable?
Actually, is see the bug you're outlining here as a feature. I
definitely think that in the absence of any author provided
information, the UA should present the quotations in a fashion
accustomed to the user. We're talking here about stylistic
conventions. It doesn't matter if a document authored by an author who
spent no effort on the styling of quotations ends up presented to a
user in a locale dependent fashion. That's what we should expect. If
the author wants to control the specific styling of the document then
that author needs to specify at least a document language and even
better provide a document linked or embedded stylesheet to specify
precisely the presentation style for quotations.
In many ways, the styling of quotations is one of those styling areas
that is ripe for user stylesheet. In other words it is an area where
the users/readers/consumers of content should care more about the
convention adhered to than the author/composer/producer of that
content. Falling back on the locale of the user then makes perfect
sense. Defaulting to my locale here in the US for a user in Nepal
makes no sense at all to me. In some cases it's possible the user
might even be completely unfamiliar with the US quotation styling
convention, but would be familiar with the local styling convention.
In any event, whether for quotation styles or otherwise, with the
advent of user stylesheets, no author can have final control over how
their content is presented (and to me that's a good thing and
something for authors to be kept aware about to improve device
independent authoring).
Take care,
Rob
Received on Friday, 31 October 2008 18:34:45 UTC