>>>>> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:
Georg> Updated and tested patch attached. Since getStatus and
Georg> doDispatch should handle the same lfuns I also implemented the
Georg> missing doDispatch methods. This is going in unless I get a no.
Minor nit:
+void MathSplitInset::doDispatch(LCursor & cur, FuncRequest & cmd)
+{
+ switch (cmd.action) {
+ case LFUN_TABULAR_FEATURE: {
+ recordUndo(cur);
+ string const s = cmd.argument;
+ if (s == "add-vline-left" || s == "add-vline-right") {
+ cur.undispatched();
+ break;
+ }
+ }
+ default:
+ MathGridInset::doDispatch(cur, cmd);
+ }
Why is reCordUndo needed? I understand you copied the code, but I do
not see any need for it...
Also, I am not sure I agree that deDispatch needs to have this entry.
In theory, doDispatch should never be invoked if getStatus does not
want to. We could maybe even assert on that in InsetBase::dispatch.
Or do you have reasons to think this is useful?
JMarc
