Re: [Denemo-devel] Denemo Hooks

[Richard Shann]

On Mon, 2011-06-27 at 22:53 +0200, R. Mattes wrote:
> On Mon, 27 Jun 2011 12:09:35 +0100, Richard Shann wrote
> > I hit the Eb/D# problem again while entering notes from my MIDI
> > controller, and that has galvanized me into action.
> > I propose to add a hook to the Key Signature settings (perhaps to *all*
> > denemo built-in commands?) and use it to adjust the range of enharmonics
> > when the keysignature is set or a new one inserted. (More is needed but
> > this email is about the hook technology).
> 
> Hmm, I'm not totally convinced that looking at the key signature would
> be a good overall strategy to decide enhamonic selection. I.e. in the
> key of A minor, how would you select between C# (often in cadences on
> the final with major third - so called 'picardian third') and Db (as
> the minor six above the forth degree in the bass - neapolitan sixt)? I
> could come quite a lot of similar probematic cases. I _do_ agree that
> it would be nice to have some way of controlling enharmonic selection,
> and selection by key signature would be a good default candidate. 
Yes, however not to interfere with manual setting (and not to get *too*
fussy) I intend to set it only initially and on keysig insertion/setting
and to store it with the movement.
>  For
> my use-cases, being able to manually switch the "tonal center" would
> be fine.
well, we have that, of course.
> 
> > So I have built the captain-hook branch and run the test from the
> > README. It is all working straight out of the box. Great Work!
> 
> Good to hear. Just one warning: I'm not shure the function names are
> written in stone yet. I'm open for suggestions ....
They seemed good to me - no mention of being just for MIDI-in, which the
README seems to hint.
> 
> > Now I have to get my head round it:)
> > I think I have to write a call to denemo-run-hook (using a scm_applyxxx
> > I think it is) at the point where the set keysignature command is about
> > to be run (and perhaps another call to denemo-run-hook after doing 
> > the keysig change, and/or perhaps not run the keysig change command 
> > if the hook returns a #f?). The parameters to denemo-run-hook would 
> > be some hook list and the value of keysig being set. So, what is a 
> > good design here for the hook list? One per built-in command? Is the 
> > overhead here tiny if the hook list is not set?
> 
> Again, I'm not sure a hook would be a good solution for this
> problem. What would a hook provide that a simple function wouldn't? 
I had just come to that conclusion :) while doing the setting the tuning
work.
But if we were providing a callback for every Denemo command, would that
naturally use your scheme?

Richard


>  I
> think Common Lisps generic functions would be ideal for such tasks:
> one could define :around, :before or :after methods. Or Emacs-like
> advice functions. Maybe GOOPS supports CLOS-style methods. It has been
> quite a while since I last looked at GOOPS. Back then the C interface
> was just non-existant.
> 
>  Cheers, RalfD
> 
> > Richard
> 
> 
> --
> R. Mattes -
> Hochschule fuer Musik Freiburg
> address@hidden
>