17 Apr
2007
17 Apr
'07
11:47 a.m.
Taco Hoekwater
David Kastrup wrote:
If there is not, it would be easy to implement, wouldn't it? Implementation would not be hard, afaict.
Skips require reference counting. That's quite a bookkeeping nuisance.
That is possible, but not required. We can just as easy store the entire glue as fields inside the math nodes. For now, we should focus on whether or not a move from dimen to skip would have unwanted side-effects.
I think that the mathsurround does not actually make it into the surrounding horizontal list explicitly. This might make it hard to let it partake in glue calculations properly. -- David Kastrup