Re: Page-level and block-level attributes

Hi,

> If I look at reference sources as zvon.org and the excellent book of
>  Dave Pawson, it is stated that spaces and indents may also be used on
> page-level objects as fo:simple-page-master and fo:region-body. The only
> formatter that implements this is XSL-formatter from Antenna House. FOP
> and XEP don't implement it this way (XEP even stops).

XEP stops because in the DTD, we have allowed only margin-* stuff  inside
the repective ATTLISTs. A note on this is present in the comments inside
the DTD, see http://www.renderx.com/Tests/validator/fo.dtd.html.
If you prefer XEP to silently ignore these attributes and go ahead, you
can turn validation off or modify the DTD (located in etc/fo.dtd inside
XEP's installation).

However, there are many problems with using spaces to set margins on
page-reference-area and region-areas. Spaces are a compound type, with
components designed to express stacking constraints. Top-level reference
areas can hardly have any constraint, so this machinery is pretty useless
but sometimes hard to interpret. Just think about .conditionality: region-areas
that are first in their respective page-reference areas should have
conditionality explicitly set to "retain", shouldn't they? so a plain 
space-before="1in" would have no effect on fo:region body :-).

> Who is wrong? Standard, references, or software?

I believe that the question is purely scholastic. Using spaces
on page-level areas is not well-defined and therefore not safe;
moreover, I cannot imagine a scenario when this feature may
turn useful. Even though I am not 100% convinced about its
legitimateness or lack thereof, I still believe that keeping
users away from these constructs is a Good Thing :-).

Best regards,

Nikolai Grigoriev
RenderX

Received on Thursday, 12 September 2002 13:46:06 UTC