Those \unexpanded 's are not necessary, sorry.

On Wed, Dec 7, 2011 at 4:42 PM, Peter Park Nelson <
peter.park.nelson@gmail.com> wrote:

> Update:
>
> I don't have a solution yet, but here is my attempt to figure out where
> this problem comes from. I may be (probably am) wrong; I am new to ConTeXt
> and not an expert on its internals. But I'll toss this out to see if anyone
> is interested.
>
> When \startcolumnset is called, it seems to take a "snapshot" of the state
> of the layout (via setuplayout). However, layout can change, even without
> \setuplayout being explicitly called again, as in the case of header
> state=high (which modifies the layout of  the first page only). By visual
> inspection, it looks like the columnset output routine continues to use the
> page dimensions from the first page on subsequent pages. Since the first
> page has no header space, height of the text box on the first page will be
> larger than on subsequent pages. Therefore, applying the first page height
> on subsequent pages will result in text that runs over the footer area.
>
> To see if a new call to \startcolumnset would fix things, I tried to find
> a way to automatically do \startcolumnset again at the beginning of the
> second page. A crude hack, using the startpostponing mechanism:
>
> \startpostponing[2] % do this block on the second page
> \unexpanded\stopcolumnset\unexpanded\startcolumnset[name] % manually end
> the current columnset and then start it again.
> \stoppostponing
>
> This unfortunately results in a blank page after the first page; however,
> the rest of the pages after that blank page have correct columnsets and
> footers.
>
> So a real solution to this problem could perhaps be done in two different
> ways -- the difference being where in the ConTeXt code it is implemented:
> A. The columnset output routine "listens" for changes in the layout state
> and reloads or adjusts itself when layout state changes; or,
> B. Any change in layout state passed to the output routine triggers a
> re-call of the \startcolumnset macro.
>
> These are probably oversimplified, but I hope someone who understands the
> columnset output routine can see what I am trying to get at.
> --
> Peter Park Nelson
> peter.park.nelson@gmail.com
>



-- 
Peter Park Nelson
peter.park.nelson@gmail.com