Re: [office] Updated proposal for lines (underlining, crossing-out,overline)

[Michael Brauer <michael.brauer@sun.com>]

Hi,

David Faure wrote:
> As discussed, here's an updated proposal for the underlining/crossing-out-overline
> specification, more based on CSS3.
> 
> text-underline-type: none | single | double
> text-underline-style: none | solid | dotted | dashed | long-dash | dot-dash | dot-dot-dash | wave
> text-underline-width: auto | thick
>    or the full CSS3 spec: auto | <normal> | <number> | <length> | <percentage> | thin | medium | thick
>    (does any office suite need so much precision on the line widths?)
> text-underline-color already defined (3.10.23)
> text-underline-mode: continuous | skip-white-space
> 
> Notes:
> * text-underline-type isn't in CSS3, I have kept it separated so that
>   "double dotted" is a possible combination, I feel this is more flexible than
>   the CSS3 mechanism of having "double" in text-underline-style

Compared to adding values like "double-dotted" to text-underline-style, 
this two-attribute approach has the advantage that all double lines are 
interpreted as single lines automatically if text-underline-type is 
ignored. This seems to be reasonable and simplifies transformations to 
CSS3 for all the double line styles that CSS3 does not support. The 
disadvantage of this solutions is that a solid double line will appear 
as single line in CSS3 if text-underline-style is ignored.

If the double line styles are added to text-underline-style, the double 
line would work in CSS3 without any transformation, but all other double 
line styles would either disappear in CSS3 or would be displayed as 
single lines probably.

 From the office application view, I personally prefer the two-attribute 
solution, but I'm not sure which solution is better from the CSS3 
transformation point of view.


> * the "auto" value for text-underline-width allows "double" underline to take
>   more space than "single" underline, says the CSS3 specification. I might be
>   a better naming for the default width. Thick would be what we call bold.

The "auto" width is also allowed to consider the font size, but "thick" 
unfortunately isn't. In OpenOffice.org and at least one other 
application, a "bold" line considers the font size as well. This means 
that "thick" will be interpreted differently by CSS3 than by (some) 
office applications. For this reason, we from my point of view might 
consider to use the value "thin" instead of "auto" for non bold lines. I 
personally do not see requirement for other widths.

> * text-underline-mode would replace fo:score-spaces (3.10.25).
> The reason is that it's IMHO easier to have all settings relating to underline
> as text-underline-*, and it separates "should I underline spaces" from
> "should I cross out spaces". We have this distinction in KWord already,
> and at the moment we have to bundle the two into fo:score-spaces. It's a
> small difference, but I figure we should go for the most flexible approach.

To replace fo:score-spaces with text-underline-mode seems to be 
reasonable, but makes transformations to XSL-FO more difficult. CSS1/2 
doesn't support score-spaces, so transformations to CSS1/2 wouldn't be 
effected.

> * You can add small-wave to text-underline-style if you see the need for it.

I think we can/should remove this value.

> * I have added long-dash, it's not in CSS3 (but it's in OO).

It's supported by other applications as well.

> * Given that this comes from CSS (and not from XSLFO), I assume the proper
> namespace for those attributes is "style:", right? I haven't seen any "css:"
> namespace, and the possible values for text-underline-style are a little bit
> different from CSS anyway.

I agree that "style" seems to be the correct namespace. We might 
consider to add a "css" or "css3" namespace, but it seems to me that at 
least for underlining, small differences to the css3 attributes will exist.

> 
> The same can be used with overline instead of underline, for the lines over
> the text (IIRC this is possible in OO?).

No, OOo does not support overlining, but we of course can add it to the 
OASIS specification regardless whether OOo supports it or not.

> 
> For crossing out, CSS3 uses the same as above, named text-line-through-*, 
> but this is only a starting point for us, we need to add the possibility to 
> cross out with characters, which I'll let Michael look into.
> Tab stop leaders will need something very similar to the crossing out spec.
> 

Best regards

Michael