LATEX-L Archives

Mailing list for the LaTeX3 project


Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Frank Mittelbach <[log in to unmask]>
Reply To:
Mailing list for the LaTeX3 project <[log in to unmask]>
Fri, 22 Feb 2008 12:27:07 +0100
text/plain (83 lines)
as Tobias read this list through gmane I have forwarded his comments


Tobias Schlemmer writes:
 > Hi,
 > Though I don't have the right to post to the LaTeX3 mailing list I
 > sometimes read it. I write my comments to you. Feel free to post or to
 > discard them.
 > Frank Mittelbach schrieb:
 > > There is another snag with the above algorithm: after having changed to a new
 > > page setup the galley space in the new setup might have changed (if page setup
 > > means more than just float areas, e.g., the vertical size of the columns might
 > > differ or the number of columns on the page might differ). Thus that means
 > > that after the second step the galley size might be smaller or larger thus the
 > > directive might already got pushed out (which would be bad since it means we
 > > have two instable states) or (nearly equally bad) we get more space in the
 > > galley and as a result hit a later directive requiring a different page
 > > setup. 
 > > 
 > > So what would be the alternatives? Anybody any good suggestions? Here is one:
 > > 
 > >     * Use the directives to always specify what should happen on the next not
 > >       on the current page. This way we could simply wait until we typeset the
 > >       final page and pick up the directives in the text at that point thus
 > >       knowing exactly what should happen with the next page. 
 > > 
 > > One problem that I see with this one is that one isn't used to it,
 > > e.g. instead of \chapter{foo} =\thispagestyle{empty}= one would now have to
 > > say something like \nextpagestyle{empty} =\chapter{foo}=. Not being used to
 > > isn't in itself a problem (though it might be one for adoption by the
 > > users). But this also means that such directives can't be easily issues from
 > > within other commands, since in such a case the area of influence is the page
 > > on which the current point will finally end up and not the page after
 > > thereafter. Now with a command that itself starts a new page the command can
 > > push out the directive, then starts the new page, but if the command doesn't
 > > start a new page then directives can't be used. 
 > > 
 > > Now I don't know how much of a problem this is (if any) but it is something
 > > that needs to be considered as it would be a restriction if this approach is
 > > used. 
 > > 
 > > Any other suggestions how one could conceptually handle page control?
 > I didn't understand everything very well, so maybe my thoughts have been
 > posted already.
 > I think many decitions can be performed by the author. He could decide
 > if he likes to have a stable layout change on the next page or on the
 > current page if applicable. I'd like to have four commands:
 > \pagestyle            \thispagestyle
 > \fromnextpagestyle    \nextpagestyle
 > The first row acts on the page where the directive is given and the
 > other two on the next page. The first column acts on the following pages
 > too while the other only describe the style for one page.
 > I think in flip flop situations the pagestyle command must be moved
 > elsewhere for a proper layout. But this must be done by the author. So
 > I'd expect LaTeX to insert a page break immediately before \pagestyle of
 > \thispagestyle if it can't find a suitable solution.
 > I'd expect in all situations a normal float placement, as it would have
 > been without a change. If there are more than one pagestyle commands
 > given on one page, then I'd expect, that they are all tried and the best
 >    Solution will be performed. This would introduce some weighting
 > mechanism between more floats, more text, least lost space at the end of
 > the page if a pagebreak follows and first/last prefered. This weighting
 > should be user configurable. I think of measuring the badnesses of the
 > criteria serparately and performing a dot product with the (user given)
 > weighting vector.
 > Tobias