Andre Poenitz wrote: > I don't mean the find scheme, but the 'dialog' scheme. To search for a > 'real' formula one must be able to enter it in the search dialog. Which, > in turn, means that the search dialog needs to accept almost arbitrary > data instead of a simple string. Sort of 'embedded InsetText'...
I don't know if I like it so much. I'd like much better plain text + regexps. I think it's way more powerful. I don't see how well this embedded InsetText can even play with f.ex. regexp. > > What I think is a nice feature of the current scheme (it's not > > implemented yet, but it's easy to do) is to treat text like a flow, > > allowing to find things spanning at different levels. > > That's what I'd do in math: Translate search string and formual to plain > text (i.e. LaTeX) and match these. This can cover several levels. Curiosity: how do you keep the sync between the plain text position and the math position for the final cursor positioning? > We already have these decorations in the read & write methods. No need > to invent a second 'language' to do basically the same thing. > > Of course, nobody would type '\layout Subsection\nFoo' in the search > input field right now. But with an embedded InsetText there asking the > user to type in 'Foo' and format as 'Subsection' is feasible... I have think about this, but I still think that regexp searches are simpler and powerful enough for handling things like this painlessly Alfredo
