Hi Hans, many thanks for your explanation and sorry for the trouble. I created the test case that only contains three issues related to page breaks. Many thanks for the new beta, I’m afraid I’m still experiencing issues. \showframe\showgrid \setupbodyfont[pagella,15pt] \starttext \dorecurse{100}{This is a sentence. } \startitemize \dorecurse{3}{\item This is an item.\par} \stopitemize \stoptext Previous beta located two items on the first page. Latest beta locates only one. But when items are higher than two, the first item disappears from the first page in the second run. And my first sample works fine with the new beta, but not with the real book. I’ll try to get a new sample. Many thanks for your help again, Pablo. On 06/01/2014 01:51 PM, Hans Hagen wrote:
Hi Pablo,
The section breaks are a bit of a headache. In the old mechanism (also mkii) we used penalties between different section levels (different values so that there is some hierarchy).
In mkiv we have a weighted penalty/glue mechanism which has advantages but can also create side effects. The method of a few weeks ago could deal with
\section followed by 100 \subsections
i.e. no text between this. In a strict mechanism you end up with one long page fur to all the penalties. So, there was some precaution for this. However that didn't work well with content following sections that prefers breaks (read: section head followed by e.g. itemize or tabulate).
So, the newer mechanism tries to analyze the page stream which (currently) is somewhat tricky. That's why we need to iterate to an acceptable solution.
Some of these mkiv mechanisms are already a bit old and could be done better using some newer luatex trickery (but i need tome to do that).
It would be nice to have a large test case for this (sort of hard to make one as some trial and error is involved). As you already have some, best collect them for future reference.
A new beta ...
Hans