On Sat, 24 Sep 2005 03:45 am, Andreas L Delmelle wrote:
> On Sep 23, 2005, at 18:12, Andreas L Delmelle wrote:
> > <snip />
> > It almost makes it seem as if padding always needs to be specified
> > as a percentage-width, or somehow a fixed-length is always
> > interpreted as "the specified fixed-length * 100%"...?
>
> Or, more correctly: as if the fixed-length is converted to a
> percentage only to be recalculated, so that you end up with the exact
> same value...?
>
> Somehow this seems suboptimal. Let me know if I'm missing anything.
>
Andreas,

I think this is a misunderstanding. Firstly you can specify a padding 
width in any way allowed by the spec. However, if you want to get the 
computed/absolute value anywhere you need to provide a context for 
percentage evaluation as the width in question could have been 
specified by the user as a percentage. The property system will only 
use the context if the specified value is a percentage otherwise it 
will ignore it.

The case in question is tricky as the context is only available during 
layout but in this case a padding width is attempted to be evaluated 
during fo tree parsing. In the general case that can not be done as if 
the width is specified as a percentage you have to wait until layout to 
get typically the ipd on which the percentage is based. Therefore the 
whole logic of trying to determine "hasPadding" during fo tree 
construction is misplaced. This is further complicated by the 
possibility that the specified value  can be an expression containing a 
percentage (eg. 5% - 5pt). There is no way to determine if this 
expression evaluates to 0 or not until layout.

> Thanks,
>
> Andreas

HTH

Manuel

Reply via email to