Re: A quick Performer_group query

[Ralph Little]

Han-Wen Nienhuys wrote:
> 2007/10/9, Ralph Little <address@hidden>:
>   
> > Hi,
> > More strangeness.
> >
> > In Performer_group::do_announces(), there is this:
> > ==================================
> >   while (1)
> >     {
> >       performer_each (get_simple_trans_list (),
> >               &Performer::create_audio_elements);
> >
> >       if (!announce_infos_.size ())
> >     break;
> >
> >       acknowledge_audio_elements ();
> >       announce_infos_.clear ();
> >     }
> >     
>
>
> this looks fishy; better take a look at the one for grobs. The idea is that
> new objects may be created in response to acknowledged objects. Then these
> new objects have to be announced, etc.
>   
Hi,
In retrospect, I think it is "OK" in what it is trying to do.
Looking at Engraver_group::do_announces, it is clearly copied originally 
from that.

However, no performer actually derives from 
Performer::create_audio_elements() which is a NOP, so it's a bit 
pointless. I was worried that it might be a performance hit on Lilypond 
if it was unnecessarily repeating calls where there was nothing left to 
do. It appears now that in practice it does exactly nothing anyway.

If create_audio_elements is not used, perhaps it could be dropped unless 
there is a structural need for it in circumstances that have yet not 
been encountered. I do not yet have a sufficient overall feel for how it 
fits in to give an authoritative view on this.

Regards,
Ralph