Re: [PATCH] Add a way to override the part-combination decision

[Reinhold Kainhofer]

Am Montag, 20. September 2010, um 20:25:30 schrieb Han-Wen Nienhuys:
> On Mon, Sep 20, 2010 at 3:07 PM, Reinhold Kainhofer
> > My only problem now is that when I add definitions like (so that one can
> > also force other types than the pre-defined ones):
> > 
> I am not sure.  The warning comes from generic code in translator.cc.
> Since the partcombiner passes music through translation twice (once to
> determine the split list, once to do the actual typesetting), it may
> be helpful to figure out if this comes from the first or the second
> run.  Without looking more closely I'd suspect that somehow two
> different force events are delivered to the same context in the second
> pass.

Sigh, okay, the problem was that I thought it would be appropriate for the 
PartCombineForceEvent to be in the event-class part-combine-event, too. 
Unfortunately, the UnisonoEvent etc. are added by the part-combiner and the 
PartCombineForceEvent survives the part-combiner, too. Now, the 
Part_combine_engraver listens to all part-combine-events and really receives 
two (one UnisonoEvent and one PartCombineForceEvent), triggering an error in 
the listen_part_combine call, where ASSIGN_ONCE is used...

Removed PartCombineForceEvent from the part-combine-event class now.

> > And the other small issue I have: Is there any way to let a
> > music-function take a symbol (the forced part-combine strategy) or ##f
> > (to revert to automatic part-combination) as argument? That would let me
> > combine the partcombineAutomatic and partcombineApart etc. definitions
> > to one single definition.
> 
> I guess you could create a type predicate symbol-or-false? that
> accepts either a symbol or false.

I thought that would be seen as quite some hack, so I tried to find out if 
there is any other way. But weighing defining of such a function against code 
duplication, the function definition wins ;-)

Am Montag, 20. September 2010, um 21:11:57 schrieb Neil Puttock:
> On 20 September 2010 19:07, Reinhold Kainhofer <address@hidden> 
wrote:
> > Okay, I have now changed the code to use a new PartCombineForceEvent (had
> > to add a dummy translator listener to the part-combine engraver to
> > silence lilypond)
> 
> Adding the event to `unlistened-music-event-classes' will silence it
> without having to add a listener.

Ah, thanks, I missed that (I grepped for some other events to see how they are 
handled without a listener, but I neglected to check the second hit in the 
define-event-classes.scm file).

New patch is up now, which has included all comments and fixed all problems:
http://codereview.appspot.com/1698054

Cheers,
Reinhold

-- 
------------------------------------------------------------------
Reinhold Kainhofer, address@hidden, http://reinhold.kainhofer.com/
 * Financial & Actuarial Math., Vienna Univ. of Technology, Austria
 * http://www.fam.tuwien.ac.at/, DVR: 0005886
 * LilyPond, Music typesetting, http://www.lilypond.org