Re: [SVG1.1F2 LC] Comment regarding Text layout

On Fri, 25 Jun 2010 01:22:05 +0200, Alex Danilo <alex@abbra.com> wrote:

> Hi Patrick,
>
>  I don't profess to claim anything either way, but from what
> I can see the spec. section 10.7 is in error, or at least not
> specific enough.
>
>  The test is 'passed' by the WebKit browsers Safari and
> Chrome as well as the latest Opera. It also passes in Batik and
> Brand-X Canadian embedded viewer. I believe it also passes
> in the JavaME Tiny implementation though I haven't tested it
> lately.
>
>  My understanding is that the text chunking happens after
> ligature formation and so the prose in 10.7 is wrong.

Yes, I agree that 10.7 seems wrong here.

>  This also impacts the pipeline for glyph placement
> quite significantly between the 2 interpretations. But I
> think there's a clear precedent that 10.5 is OK, but 10.7
> needs changing.

I wouldn't mind to see some more tests for this.

- An equivalent test to text-text-06-t.svg but using systemfonts instead  
of svgfonts
- A test for testing ligature formation across element boundaries (e.g  
<text>f<tspan>i</tspan></text> with a 'fi' ligature)
- A test where 'letter-spacing' is varied, and whether ligatures are  
formed when a non-default value is used

My proposal is that we remove the following two sentences from 10.7:
[[ Ligatures only occur when a set of characters which might map to a  
ligature are all in the same text chunk. ]]
and
[[ As mentioned above, ligature formation should not be enabled for the  
glyphs corresponding to characters within different text chunks. ]]

 From some limited testing it seems like Batik, Opera and Webkit ignore the  
ligature 'letter-spacing' clause in 10.7 too, all still form the ligature  
even though a non-default 'letter-spacing' value is used. This raises the  
question if 'letter-spacing' should start talking about glyphs instead of  
characters, if that is the intended behavior.

Cheers
/Erik

-- 
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Received on Monday, 28 June 2010 11:54:06 UTC