Hi Ulrich,

 > > but sticking with it for a moment: golden ratio between what? (i know
 > > it is also only meant as an example, but what are the ingredients?)
 > 
 > place top of float area so that the ratio to the height of the page is
 > like the golden ratio 
 >  
 > + a   ttt  ttt
 > |     ttt  ttt
 > |          ttt
 > | b AAAAA  ttt
 > |   AAAAA  ttt
 > |   AAAAA  ttt
 > |          ttt
 > |     ttt  ttt
 > |     ttt  ttt
 > + c   ttt  ttt
 > 
 > bc/ac = golden ratio; e.g. if the height of page has 42\baselineskip
 > then b would start after line 16 (if I calculated correctly ;-)

ok so what you propose here is a golden ratio specifying the top point of the
float area (ie a special setting of my second proposed rule).

 > > so can you perhaps give some explicit pseudo specification?
 > > 
 > > \DeclareFloatArea
 > >   {  position = m  % (or t or b)
 > >     ,column   = 1  % 2 3 4 ...
 > >     ,span     = 1  % 2 3 ...
 > >    ...             % your spec
 > >   }
 > 
 > \DeclareFloatArea 
 >   {  position = m  % (or t or b)
 >     ,column   = 1  % 2 3 4 ...
 >     ,span     = 1  % 2 3 ...
 >     ,vsize    = 6\baselineskip            % vsize of area
 >     ,pos      = absolute(16\baselineskip) % vertical starting position  
 >   } 
 > 
 > > precise enough to make your picture example page come to life :-)

ok, though this needs a couple more parameters to specify how the float is
positioned within the area unless that is always supposed top to bottom (which
gives you a parameter without variation possibilities).

 > maybe a global definition which can be overridden is more useful, e.g.,

could be but that is technique ie interface design not a functionality
change/extension

 > > =========================================
 > > 
 > >  > > But what I'm after is this:
 > >  > >
 > >  > >  - assuming you have the possibility of specifying one (or more?)
 > middle 
 > >  > >  areas for floats by which I mean an area to receieve float(s)
 > where
 > >  > >  above and below there is still text
 > 
 > this should be covered by my above definition (text above
 > 16\baselineskip, float area 6\baselineskip, text after
 > 42-(16+6)\baselineskip).

as one possibility for specification (not the actual values but the concept:
which is fixed starting position with a given vertical size and one area
only). my question is what others  should/could be supported?

 > 
 > >  > <rulers>, <colour specs, including transparency>,
 > > 
 > > aren't those more kind of decorations on the area? so in other words
 > > irrelevant for placement (other than the decorative elements might
 > need space)
 > > or do i miss something.
 > 
 > yes, that's true but where do you define global definitions for the
 > visual appearance? And one should always have the possibility to
 > override some specs, e.g. in a special case one might not want top
 > rules.

sure. all those could and should eventually become part of an area
specification globally with overwrites or individually. what I mean is as far
as float placement is concerned they play no role so for the discussion I'm
currently not concerned about them.

 > > can you explain what you mean by "including transparency"?
 > 
 > You should be able to define background colour|pictures, rules (top,
 > bottom, box, coloured), transparency for background colour|pictures and
 > maybe even a gradient colour. Plus a command to override the global
 > definition in a special case.

ok so it is what I thought you meant

 >  
 > >     * The ratio of t1 to t2 is fixed by the design and a float AAA can be
 > >     placed into the middle position if neither t1 nor t2 become too
 > >     small. (Downside of this kind of layout might be that the positioning
 > >     of the floats drastically varies from page to page.)
 > > 
 > >     * The end position of t1 is fixed (vertically) so that a middle float
 > >     always starts on the same point on a page. Further restriction then
 > >     that t2 is not getting smaller as a certain value.
 > > 
 > >     * The starting starting position of t2 is fixed so that the bottom of
 > >     the middle floats always appear on the same vertical position on the
 > >     page, again with some further restrictions to the size of t1 this
 > >     time.
 > 
 > I thought this was what I'd proposed.

yes, I think so, you proposed option 2. But what about:

 > >     * ...other ideas...

is there any kind of reasonable rule set that is not covered by my initial
variations and should perhaps be supported as well?
 > 
 > sorry if my explanations are too cryptic.

not at all, it is a difficult topic after all.

frank