Frank Mittelbach <[log in to unmask]> writes: > Marcel Oliver writes: > > > What I would really like to see at the user level is the possibility > > of relative sectioning commands. > > having read this far i thought that you meant something like this: > > \begin{head}{Main heading} > > text text text > > \begin{head}{Sub heading} > text text > \end{head} [snip] So nesting would determine the level? Or perhaps one could simply descend a level with \downsection{heading} (no explicit container for section contents as with LaTeX \section) and go up a level with \upsection. Then if the value of \documentclass were, for example, "article", a "\downsection" at the top of the body would be the same as "\section". And at some point when deeply nested it would be possible to get out simply with an absolute command such as old-fashioned \subsection without risking an incorrect number of \end's. If section-like units are going to have explicit containment boundaries, then I see a sense only in absolute names and I would suggest upper casing the traditional names for this. That is, \begin{Subsection}{heading} ... \end{Subsection}, but I think there are relatively few cases where explicit containment is actually helpful. (Still I have it modeled for optional use in gellmu.dtd; I seldom use it since I'm lazy. The language definition in the dtd causes sgml endtags to be placed correctly in almost every case (perhaps every case). An OMITTAG feature available in SGML, not in XML; once it's there, one can down-translate to XML if desired.) > > One requirement should be that one > > can mix absolute and relative sectioning in a natural way. To explain > > but to be honnest i don't think that a concept of mixing absolute names with > relative names via a label/ref mechanism is really helpful. The labels could only be auxiliary as now. One keeps track of the level. Absolute names have absolute level values depending on the value of \documentclass, while \downsection increments the level by one and \upsection does the reverse. > I agree that it is a bit of a pain to change a heading tree in a LaTeX > document at the moment but something like the above seems to mak a document > completely uncomprehensible (if used on a larger scale). So in my opinion it > is better to either accept that changing the names is necessary if one > changes the structure or to use a real relative naming scheme (as outlined > above) in the first place. --- or perhaps i'm just getting too old for > revolutionary changes? (No! You certainly must be younger than I am. :-) ) What about a user-level way to tweek the section level counter? (Perhaps a new section level counter, not stepping on anything already there.) -- Bill