On Sun, Apr 10, 2005 at 06:48:33PM +0200, Georg Baum wrote: > Am Sonntag, 10. April 2005 16:22 schrieb Martin Vermeer: > > On Sun, Apr 10, 2005 at 12:32:19PM +0200, Helge Hafting wrote: > > > I wonder if this was the right approach - or would a special > > > case in pasting be better? > > > > > > Helge Hafting > > > > I am afraid I must apologize for underestimating this. It required > > considerably more doings to get it all right. The knobs and buttons were > > all there, they just had to be found and put to use. This is actually > > C++ at its best ;-9 > > > > I do believe that we should have "read" methods in all the insets to > > replace the monolithic math parser. How to do that, is another question. > > I would prefer that too, but this is _very_ difficult if not impossible as > long as math is stored as TeX. TeX is unparsable, even the subset we use > is difficult enough to parse. IMHO the current approach is not that bad. > > > See patch. > > Too confusing IMHO. I agree with Helge and would rather implement > MathRootInset::doDispatch() and handle LFUN_PASTE there. This is less > code duplication than with this patch. > > > Georg >
If you can do it more efficiently, why not. I don't like special casing, but if you can do it in MathRootInset::doDispatch() it would even be clean. One reason I didn't consider this (not recognize what was meant, actually) was my experience in implementing insert-around-selection for font and deco insets (or general nestinsets), which I did in LFUN_INSERT_MATH, not LFUN_PASTE. Actually why not implement an LFUN_INSERT_MATH in MathRootInset? See math_nestinset.C for a role model. (Note: cur.handleNest takes a second cell number arg, see cursor.h!) - Martin
pgp9pvervE4bi.pgp
Description: PGP signature
