[Rd] mtext adj= wrong with several las= (PR#7188)

[Uwe Ligges]

Peter Dalgaard wrote:

> Paul Murrell <p.murrell at auckland.ac.nz> writes:
> 
> 
>>I think the ideal would be to set padj automatically for multi-line
>>text as well as single-line text (the current behaviour tries to set
>>padj automatically for single-line text;  it's just not very good at
>>it :).
>>But whatever first step is good;  then we can more easily discuss
>>actual behaviour and code examples.
> 
> 
> A somewhat uninformed conjecture: You'll need to look at the TeXbook
> again. Think of it like this: A single line of text forms a box that
> has a top, a bottom, and a baseline, and you can align on either or on
> some fraction between top and bottom. Multiple lines form a box too,
> but you need a way to specify whether the baseline of the entire box
> is defined by the top or the bottom line inside of it. In addition,
> you need to specify alignment inside the box. Basically, we need a
> more structured approach than a bunch of options that change meaning
> depending on the setting of other options.

I think we already have that (more or less), and we don't want to 
re-implement TeX, I hope (in this case I'm the wrong person given my 
capabilities ;-)).
There is a top and a bottom and the bottom line defined as a baseline.
Then we have (speaking about mtext() now) "adj" that controls the 
right/left adjustment in reading direction. What we do not have is 
adjustment in the perpendicular reading direction - it's simply a fixed 
value that can only be controlled using "at".
All the stuff does not change meaning depending on other options, only 
the defaults are sometimes changing (in order not be be completely ugly).
My aim is (a) to provide the missing argument and (b) to set more 
sensible defaults than that one we already have. It's not that easy 
doing it without breaking code.

Uwe