That would be just a matter of associating a value with the keep.  The 
main thing about the approach I outlined is to provide a structure which 
allows keeps (and stacking constraints) to be expressed more readily.

Incidentally, speaking of min/opt/max values, my feeling is that, in a 
look-ahead model, the initial layout should use optimum values only. 
 Then, only if some layout infelicity or contradiction occurs, the 
layout manager, wherever that sits in the process hierarchy, will start 
probing the galleys for possible layouts in the minimum ior maximum 
directions.  Which to try first will probably be heuristically 
determined, based on the type of layout errors which can occur.  One of 
the variables here will be the "strength" of keeps, but, as far as I can 
tell, the weakest keep will only be discarded after the options implicit 
in min/opt/max have been exhausted.  I haven't given any thought to any 
of this, though.


Keiron Liddle wrote:

> Does this take into account the keep values that have the range: 0..9, 
> always. It is not possible to simply say that a given group of keeps 
> should stay together, you need to find the lowest keep value closest 
> to the optimum spacing position and within minimum.

To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]

Reply via email to