Re: Text properties slow down XEmacs

Hrvoje> Jan Vroonhof <vroonhof(a)math.ethz.ch&gt; writes: Hrvoje> Not really (although it may look that way at the start). Hrvoje> The font-lock code wants to apply certain properties to Hrvoje> parts of the buffer, merging or breaking the extents as Hrvoje> needed. This is exactly what the text-properties are for. SJT> Could you be more specific? The documentation says something quite SJT> different: text-properties are for attaching properties to characters SJT> rather than intervals. That's exactly what font-lock is supposed to do: attaching faces to characters matched by specific (parts of) font-lock keywords. SJT> That seems like exactly the _wrong_ thing to do with strings and SJT> comments, both of which should be locked up in an unbreakable SJT> (from the point of view of font-lock) extent once identified; There are not such things like unbreakable extents in font-lock: parts of previously fontified characters can get additional/new face properties, etc. In fact, two extents are "glumped" if appropriate, e.g., for a TeX sequence like \alpha\beta. SJT> the individual characters in them should never be considered SJT> "locally". ??? SJT> OTOH, I don't see why splitting an extent or merging neighboring SJT> extents (which should be constant time, with relatively high overhead SJT> I suppose) should be more efficient than iterating over a (possibly SJT> lengthy) interval of characters (linear time but possibly lower SJT> overhead, depending on the implementation, although I find it hard to SJT> see why it would be necessarily much less than for extents). This probably depends: the display is faster, the fontification not. - Christoph