W3C home > Mailing lists > Public > public-css-testsuite@w3.org > November 2007

Re: Questions about length unit tests

From: Dan Kennedy <danielk1977@gmail.com>
Date: Fri, 2 Nov 2007 12:01:30 +0700
Message-ID: <ee8520a50711012201t60d7334ag559db340aa5dd945@mail.gmail.com>
To: public-css-testsuite@w3.org

On 11/2/07, L. David Baron <dbaron@dbaron.org> wrote:
>
> On Wednesday 2007-10-31 18:35 +0700, Dan Kennedy wrote:
> > Checked out a fresh copy today. First question is about the test:
> >
> >   t040302-c61-ex-len-00-b-a.htm
> >
> > what encoding should the UA assume this test uses?
>
> Should we consider adding
>     <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
> to the HTML version of the tests so that they work when downloaded?
> (I'm assuming we're sending it as an HTTP header, at least?  Can't
> check right now, since I'm writing this offline.)

Right, the HTTP header does specify utf-8. I think it would be a good
idea to include a <meta> element in the document too. Hv3 assumes
utf-8, but major browsers seem to assume 8859 when loading a file://
URI.

One reason I got so confused is that t040302-c61-ex-len-00-b-a.htm
appears to pass in current versions of major browsers on linux if
loaded with 8859 encoding. This is because they get both the "test"
and "control" blocks wrong in this case - the test block is wrong because
the metrics of the Ahem font are not being interpreted correctly and
the control block is wrong because the wrong glyph is selected due to
the 8859 encoding. Unfortunately, the two "wrong" renderings are identical,
so to the casual observer the test appears to pass!

Regards,
Dan.
Received on Friday, 2 November 2007 05:01:47 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:51:55 GMT