[Pseudo-patch] Add support for HairpinText indications.

Add support for HairpinText indications.

This first attempt, although unsuccessful, aims to
implement native, tweakable text indications that
are sometimes found above or below Hairpin objects.

(See LSR snippet #233 for an example.)

[Valentin Villenave, 2010-12-07 23:32:46]

Greetings everybody,

I am currently trying to understand how Grobs are created and handled, and
here's my first attempt at fiddling with them. Needless to say, it doesn't work
(exits with a segfault) but at least it doesn't seem to break existing things.

My eventual goal would be to create a new HairpinText grob that would be printed
above or below hairpin crescendos, a bit like
http://lsr.dsi.unimi.it/LSR/Item?id=233 but in a native, Grob-ish way.

For that purpose, I've created a new 'mixed span-type (I think I got this sort
of covered now) and a new Grob (which is, for now, pure nonsense).  I could
really use some pointers (no pun intended) as to where to go from this point,
assuming my approach is correct in the first place, and this feature is worth
implementing at all (which can be argued upon).

Anyways, please have a look and let me know! Thanks!

[Neil Puttock, 2010-12-08 01:09:38]

Hi Valentin,

I've added a few comments which I hope you'll find helpful.

Cheers,
Neil

http://codereview.appspot.com/3509041/diff/3001/lily/new-dynamic-engraver.cc
File lily/new-dynamic-engraver.cc (right):

http://codereview.appspot.com/3509041/diff/3001/lily/new-dynamic-engraver.cc#...
lily/new-dynamic-engraver.cc:147: hairpin_text_ = make_item ("HairpinText",
hairpin_text_->self_scm ());
I suspect this is part of where your segfault comes from: you can't make the
cause of a grob the grob itself :)

make_item ("HairpinText", current_span_event_->self_scm ());

You need to decide what kind of grob you're creating; though you've defined
HairpinText as an Item, some of the default settings in define-grobs.scm are for
a Spanner (which will probably cause you more segfaulting issues).

Since you want to centre the text between the bounds of the Hairpin, it probably
makes more sense to define HairpinText as a spanner.

[Valentin Villenave, 2010-12-08 08:09:14]

[I should probably subscribe the codereview mail address to Frogs- and move the
discussion there.]

Thanks Neil! I've now tried to make a proper Spanner, which raises more
questions of course (namely, how to properly start and finish it): right now
there's no segfault anymore, but a "must have Item for spanner bound" error.

I'll keep looking :-)

Thanks!

http://codereview.appspot.com/3509041/diff/3001/lily/new-dynamic-engraver.cc
File lily/new-dynamic-engraver.cc (right):

http://codereview.appspot.com/3509041/diff/3001/lily/new-dynamic-engraver.cc#...
lily/new-dynamic-engraver.cc:147: hairpin_text_ = make_item ("HairpinText",
hairpin_text_->self_scm ());
On 2010/12/08 01:09:38, Neil Puttock wrote:
> I suspect this is part of where your segfault comes from: you can't make the
> cause of a grob the grob itself :)

Interesting.  I hadn't understood that the second argument was the grob cause,
but that makes sense. (I guess that's why many grobs are created with make_item
("FooBar", SCM_EOL)?

> make_item ("HairpinText", current_span_event_->self_scm ());
> 
> You need to decide what kind of grob you're creating; though you've defined
> HairpinText as an Item, some of the default settings in define-grobs.scm are
for
> a Spanner (which will probably cause you more segfaulting issues).

Yes.  As you've noticed, it was an ugly mashup of various properties found here
and there :-)

> Since you want to centre the text between the bounds of the Hairpin, it
probably
> makes more sense to define HairpinText as a spanner.

Done. I also considered the possibility of using a Scheme procedure as a
translator, but it seemed somehow uncanonical.

[Neil Puttock, 2010-12-12 00:07:03]

On 8 December 2010 08:09,  <v.villenave@gmail.com> wrote:
> [I should probably subscribe the codereview mail address to Frogs- and
> move the discussion there.]

Sounds sensible to me.

> Thanks Neil! I've now tried to make a proper Spanner, which raises more
> questions of course (namely, how to properly start and finish it): right
> now there's no segfault anymore, but a "must have Item for spanner
> bound" error.

Did you notice which grob's missing the bound, though?  It's not what
you'd expect (i.e., the HairpinText grob :)

(hint: the code you've added for HairpinText prevents a Hairpin from
being created)

> Interesting.  I hadn't understood that the second argument was the grob
> cause, but that makes sense. (I guess that's why many grobs are created
> with make_item ("FooBar", SCM_EOL)?

Not all grobs are triggered by events or other grobs, so there'll
always be some with empty causes.

> Done. I also considered the possibility of using a Scheme procedure as a
> translator, but it seemed somehow uncanonical.

Actually, I think you'd find it easier to create HairpinText in a
separate engraver (at least until you're satisfied it's working
properly); this you could prototype in scheme (since you're unlikely
to need any C++ specific code).

Cheers,
Neil