Josà AbÃlio Oliveira Matos <[EMAIL PROTECTED]> writes: > > On Thu, Oct 14, 2004 at 11:52:50PM +0200, Andreas Vox wrote: > > > > Something like a function "StringList getPossibleLayouts(LyXText & , > > LCursor)" > > which would compare the current sequence of paragraphs around the cursor > > position with allowed DTD productions. > > This would be on the fly support. It would be could but there other things > that we can do first. > > > A variant of this function could be used to check if a depth change is > > allowed > > for a given paragraph. > > > > Questions: > > * What do you think of it ? > > The first time we talked about this was 1997.
Great! Where is the code? ;-) > > > * If this functions exists, would it be difficult to use it to prune > > the lists of layouts ? > > To make it dinamically, you mean? Exactly. > > Are there problems with some frontends ? > > Jean-Marc can answer that. > > > If you like the idea, I would think about an algorithm to implement > > this function. > > I like the idea. Note that something easier that can be done now is to > have a mode to validate the document. And then have a list of errors point > to the real location in the lyx file. > > Something like what we have now in the latex backend. Finding the first error would be easy. Avoiding followup errrrs and only reporting genuine additional errors is a challenge. In this way the dynamical approach is easier (just forbid any first error and you wont have more than one at any time -- well, of course there is the paste action ... And probably other ways to mess up the structure (includes? dtd changes?) ) I just brought it up at this time because it appears to be closely linked to 1. translation LyXBuffer -> XML 2. Deeper/depth semantics . So maybe it makes more sense to aim at a concept which can solve all three problems. Ciao /Andreas