Re: Whitespace, punctuation and :first-letter

On Thu, Oct 07, 2010 at 08:26:41AM -0700, Simon Fraser wrote:

> I made a change last night in WebKit to have a run of
> <whitespace><punc>character<whitespace><punc> be treated as :first-letter
> (where <whitespace> includes non-breaking spaces, and <punc> is the
> categories noted in the spec).
> 
> An issue I did not handle is how newlines in that whitespace are treated,
> particularly if "white-space: pre" is applied.

I think Simon means "A bug I did not correct", but I'll be explicit just in case: 
the spec seems quite clear that :first-letter must only select the first letter
on the first line.  (E.g. "The first letter must occur on the first formatted
line." and the example in the same paragraph.)

The spec is not clear whether any selected following punctuation must be on the
first line or not, and a literal reading might be that the following
punctuation is not required to be on the first line.

For that matter, a literal reading of the spec is that all following
punctuation should be included, whether or not preceded by non-selected
characters.

I propose that it be made clear that it's restricted to the first line, and
only include *immediately* following punctuation (or punctuation and Unicode
whitespace characters, if we so decide; but presumably excluding any whitespace
after the last selected non-whitespace character).

I've already indicated some misgivings about including following punctuation,
in that Le Monde seems not to include following apostrophes in first-letter
styling.  A hypothetical (though I expect rare) example where we really don't
want to include following punctuation after space is if that following
punctuation is an opening bracket of some kind.  A hyphen or dash should also
not be selected (though selecting would not as harmful as in the opening
parenthesis example).  This counts as a (weak, assuming the harmful cases are
rare) argument not to allow whitespace after the first letter/digit when
deciding how much to select.

I would expect existing practice not to include the closing ‘»’ in first-letter
styling in languages that space their quotation marks.  Take that with a grain
of salt given that I'm an English speaker, but it does seem that stylized huge
guillemets in French (when quoting a whole speech for example) are more likely
to be on a separate line than the corresponding huge quotation marks one
sometimes sees in English (which are more likely to be placed such that the
block of quoted text appears roughly where a same-size-as-the-quotation-marks
letter would appear).

pjrm.

Received on Thursday, 7 October 2010 21:57:32 UTC