On Wed, Aug 08, 2007 at 12:00:08PM +0100, Chris Bowditch wrote: > Vincent Hennebert wrote: > >> Hi all, >> There???s something quite strange in the layout code that I suspect comes >> from the pre-Knuth era. > > Yes I think you are right. > >> AbstractLM defines a reset method that among other things calls >> resetPosition on children LMs. If you search for references to this reset >> method, you find out that it is only called within resetPosition methods. >> But those resetPosition methods are themselves only called by the reset >> method! >> Unless I missed something those methods aren???t useful anymore and could >> be removed. Anyone against? It???s not like the layout managers needed to >> be cleaned up... > > IIRC, those methods are intended as a means of allowing the IPD to change. > Currently the LMs dont allow the IPD to change mid page sequence, but if > the reset methods are properly implemented then if a change in IPD is > detected then reset would be called on the PageSequenceLM. Of course, I > don't believe the reset methods are properly implemented for the Knuth > model, so you could remove them now and start from scratch when > implementing support for changing IPD.
In the best-fit method the code always went one node too far, and then reset itself to the best node using these reset methods. Indeed, they do not seem useful anymore. Simon -- Simon Pepping home page: http://www.leverkruid.eu
