[vh]box size allocation question (again).

[johannes nada kth se (Johannes Keukelaar)]

Hi All,

A week ago or so, I posted this question, which, if I remember correctly, 
hasn't been answered yet. I still wonder:

It seems that, in 1.0.2 at least, the size allocation of a vbox (and presumably 
an hbox as well) tries to divide the size it's given equally over all its 
children (rounding down), and gives the remainder to the last one. This means 
that the last child can be significantly larger (number of children minus one 
pixels) than the other ones. See for example 
http://www.nada.kth.se/~johannes/vbox.gif for an example. This, IMHO, looks 
fairly ugly. Mind you, I'm not sure if any other solution would look any 
better. I'm just wondering: Is this a conscious design choice? Wouldn't 
'dithering' the sizes (i.e. do the computations in floating point, then round 
down) look better? Then the size difference is at most one pixel...

Regards,

Johannes.
--
"You cannot save time -- you only choose how to spend it." -- Buck Tilton

<insert funny description here>