Thank you for your suggestions, Lou & Frederik.
Even if TEI coordinates and CSS positioning are incompatible, that 
doesn't necessarily mean their combination should be discouraged. It 
just needs to be made clear that they refer to different positioning 
systems, so that if processed, they are processed independently.
Either way, the Guidelines need to address this. But isn't there a major 
overhaul of these Guidelines sections underway anyway?

Martin de la Iglesia
Metadata and Data Conversion

Georg-August-Universität Göttingen
Göttingen State and University Library
D-37073 Göttingen

Papendiek 14 (Historical Building, Room 1.602)
+49 551 39-14070 (Tel.)
+49 551 39-3468 (Fax)

[log in to unmask]

Am 06.11.2015 um 09:54 schrieb Frederik Elwert:
> Dear Martin, dear all,
> I think that’s an interesting question.
> Am 06.11.2015 um 09:13 schrieb Martin de la Iglesia:
>> last week at the SIG Text & Graphics meeting we discussed, among other
>> things, an issue related to the recent addition of @points to <surface>
>> (<>): in several
>> places, the Guidelines suggest to make use of the CSS box model (e.g.
>> "margin-top: 0.5em" in Rendition) or other means of
>> horizontal/vertical positioning with CSS (e.g. "text-align: center" in
>> the <rendition> example). What happens if such horizontal/vertical
>> positioning code occurs within a non-rectangular writing area? Consider
>> this example:
>> <surface ulx="0" uly="0" lrx="3" lry="3">
>>     <zone points="1,1 2,3 3,2">
>>        <surface>
>>           <zone style="margin-top:0.5; margin-left:0.5">
>>              <line>some text here</line>
>>           </zone>
>>        </surface>
>>     </zone>
>> </surface>
>> What would this code mean? Is the text inside (i.e. "margin-top:0.5;
>> margin-left:0.5" is meant to be relative to the physical borders of the
>> parent <surface>) or outside (i.e. relative to the grid specified in the
>> grandparent <surface>) the triangle?
> I think it would make sense to have a look at CSS shapes:
> One should probably
> discourage the use of margin-* and padding-* in combination with
> @points, and rather use shape-padding. @points would probably roughly
> translate to CSS shape-inside: polygon().
>> Another problem is that you'd normally have to specify units in CSS
>> (i.e. the code above is not valid CSS), whereas you can't specify the
>> unit for @points, @ulx etc. How do I say both the CSS and coordinate
>> attributes are meant to use the same unit, say, centimetres? How do I
>> say they're not?
> I don’t remember the details, but I think SVG has a way of solving that.
> They also use unit-less coordinates but allow for CSS as well.
>> Neither is it clear to me how the new surface/@points can actually be
>> used. The Guidelines say, coordinate attributes in <surface> specify the
>> grid, while coordinate attributes in <zone> specify a position within
>> such a grid. Can I use surface/@points to specify a position on a grid?
>> Does the grid need to be specified in the same <surface>, or in an
>> ancestor <surface>?
> I see your point, but without any answer.
> Best,
> Frederik