Allow \shape to tweak music, swap its arguments

Allow \shape to tweak music, swap its arguments


Also does

Run scripts/auxiliar/update-with-convert-ly.sh

Add string-or-music? predicate

[david.nalesnik, 2012-10-02 23:39:05]

LGTM

https://codereview.appspot.com/6585052/diff/1/input/regression/shape-other-cu...
File input/regression/shape-other-curves.ly (right):

https://codereview.appspot.com/6585052/diff/1/input/regression/shape-other-cu...
input/regression/shape-other-curves.ly:18: \override PhrasingSlur #'color =
#blue
I wonder if it would be helpful to alter one or two of the following
applications of the function as override to the tweak form.

https://codereview.appspot.com/6585052/diff/1/input/regression/shape-slurs.ly
File input/regression/shape-slurs.ly (right):

https://codereview.appspot.com/6585052/diff/1/input/regression/shape-slurs.ly...
input/regression/shape-slurs.ly:19: g,8 f' e d c2)
What about adding a demonstration here of the ability to modify curves beginning
at the same timestep?  So possibly insert here something like the following
(paired with the alteration I've added below):

\break
\acciaccatura b8^( cis4 cis' fis, gis)

https://codereview.appspot.com/6585052/diff/1/input/regression/shape-slurs.ly...
input/regression/shape-slurs.ly:33: g,8 f' e d c2)
\break
\shape #'((0 . 0) (-0.1 . -0.25) (0.1 . -0.25) (0 . -0.25)) Slur
  \acciaccatura b8
    -\shape #'((-0.5 . 0.5) (0 . 0.5) (0 . 0) (0 . 0)) ^( cis4 cis' fis, gis)

[janek, 2012-10-03 05:04:11]

i've skimmed over the discussion in
http://code.google.com/p/lilypond/issues/detail?id=2858 and i'm confused.

Do we want to change 
\shape GrobName #'offsets
into
\shape #'offsets GrobName
or 
\shape #'offsets 
(used as a tweak, with GrobName being guessed)?

Janek

[dak, 2012-10-03 13:46:14]

On 2012/10/03 05:04:11, janek wrote:
> i've skimmed over the discussion in
> http://code.google.com/p/lilypond/issues/detail?id=2858 and i'm confused.
> 
> Do we want to change 
> \shape GrobName #'offsets
> into
> \shape #'offsets GrobName
> or 
> \shape #'offsets 
> (used as a tweak, with GrobName being guessed)?

Yes.  It is quite valuable to be able to use this function as a tweak (for
example, if you want to meddle with tied chords which LilyPond is not
spectacularly good at), and multiple tweaks to the same music stack reasonably
well syntactically only if the tweaked music is at the end.  So the somewhat
more natural argument order when used as an override is sacrificed in order to
make \shape more versatile.

When used as a tweak, the grob name is not "guessed" but only grobs created
directly from the tweaked music event are affected, whether or not they share
the same grob name.  That is the usual behavior of \tweak.

[dak, 2012-10-03 13:57:25]

On 2012/10/02 23:39:05, david.nalesnik wrote:
> LGTM

[...]

> I wonder if it would be helpful to alter one or two of the following
> applications of the function as override to the tweak form.

[...]

> What about adding a demonstration here of the ability to modify curves
beginning
> at the same timestep?  So possibly insert here something like the following
> (paired with the alteration I've added below):

To those suggestions let me answer with the famous answer to the question "Mr
Gandhi, what do you think of Western civilization?", namely with "I would
consider it a good idea."

Had Gandhi been a programmer, his answer might have been "Patches welcome."  The
long and the short answer is that I would not know where to start.  I have not
actually considered the documentation, and I don't even know whether doing so
would not interfere with issue 2858
<URL:http://code.google.com/p/lilypond/issues/detail?id=2858> which is currently
on hold due to this pending issue.  So the best I think I can do at the moment
is to crosslink your suggestion to issue 2858 (which is about documenting
\shape) and see how and when and where Trevor thinks it should be incorporated.

[david.nalesnik, 2012-10-03 15:08:30]

On 2012/10/03 13:57:25, dak wrote:

[...]
 
> To those suggestions let me answer with the famous answer to the question "Mr
> Gandhi, what do you think of Western civilization?", namely with "I would
> consider it a good idea."

Oh, for just one zinger like that...

> So the best I think I can do at the moment
> is to crosslink your suggestion to issue 2858 (which is about documenting
> \shape) and see how and when and where Trevor thinks it should be
incorporated.

OK, thanks for doing this.  Here I was simply thinking that the reg tests should
cover as many situations as possible (within a modest length), but I suppose
that the changes I proposed aren't testing anything that isn't covered elsewhere
("Tweaks work!")

[dak, 2012-10-03 15:17:26]

On 2012/10/03 15:08:30, david.nalesnik wrote:
> On 2012/10/03 13:57:25, dak wrote:
> 
> [...]
>  
> > To those suggestions let me answer with the famous answer to the question
"Mr
> > Gandhi, what do you think of Western civilization?", namely with "I would
> > consider it a good idea."
> 
> Oh, for just one zinger like that...
> 
> > So the best I think I can do at the moment
> > is to crosslink your suggestion to issue 2858 (which is about documenting
> > \shape) and see how and when and where Trevor thinks it should be
> incorporated.
> 
> OK, thanks for doing this.  Here I was simply thinking that the reg tests
should
> cover as many situations as possible (within a modest length), but I suppose
> that the changes I proposed aren't testing anything that isn't covered
elsewhere
> ("Tweaks work!")

David, I honestly have not really looked into this at all and did not even
realize you were talking about regtests rather than documentation (I have not
touched either myself but rather let the convert-ly rule deal with swapping
argument order).  So if you consider regression test additions orthogonal to
Trevor's documentation work, you are certainly more acquainted with the code,
and I am currently immersed in parser work to make the Context.GrobName thing
fly.  So any actual git-format-patch proposals to be folded into this issue or
as a separate add-on issue would be more than welcome.  And it is never wrong to
have the regtests for one feature test all aspects of that feature, and not rely
on something else to prove that.

[david.nalesnik, 2012-10-03 15:53:26]

On 2012/10/03 15:17:26, dak wrote:

[...]

> and I am currently immersed in parser work to make the Context.GrobName thing
> fly.  

Thank you very much for working on this!  I am looking forward to being able to
fix the mess at the beginning of \alterBroken (and a general offsetting function
I hope to put up for comments at some point).

>So any actual git-format-patch proposals to be folded into this issue or
> as a separate add-on issue would be more than welcome.  And it is never wrong
to
> have the regtests for one feature test all aspects of that feature, and not
rely
> on something else to prove that.

OK, I think I'll wait till this patch and the Trevor's documentation patch is
through, and then I'll create a new issue for the regtests.  I may simply leave
these two as they are, and add a short test along the lines of, "\shape works
independently on curves beginning at the same timestep."

[janek, 2012-10-03 16:38:45]

On Wed, Oct 3, 2012 at 3:46 PM,  <dak@gnu.org> wrote:
> On 2012/10/03 05:04:11, janek wrote:
>>
>> i've skimmed over the discussion in
>> http://code.google.com/p/lilypond/issues/detail?id=2858
>> and i'm confused. Do we want to change
>> \shape GrobName #'offsets
>> into
>> \shape #'offsets GrobName
>> or
>> \shape #'offsets
>> (used as a tweak, with GrobName being guessed)?
>
> Yes.

uh, i ususally find that it's a bad idea to answer "yes" to a question
that goes like "do we want X or Y?"
:)

> It is quite valuable to be able to use this function as a tweak

...but from this sentence i deduce that we're trying to do the second
thing, i.e. \shape will have one #'offsets argument and no argument
specifying grob name, which will be deduced from the music that
follows the command.  Right?

In that case, LGTM.

cheers,
Janek

[t.daniels_treda.co.uk, 2012-10-03 17:05:43]

david.nalesnik@gmail.com

> OK, I think I'll wait till this patch and the Trevor's documentation
> patch is through, and then I'll create a new issue for the regtests.  I
> may simply leave these two as they are, and add a short test along the
> lines of, "\shape works independently on curves beginning at the same
> timestep."

Sounds good!

Trevor

[dak, 2012-10-03 17:36:51]

On 2012/10/03 16:38:45, janek wrote:
> On Wed, Oct 3, 2012 at 3:46 PM,  <mailto:dak@gnu.org> wrote:
> > On 2012/10/03 05:04:11, janek wrote:
> >>
> >> i've skimmed over the discussion in
> >> http://code.google.com/p/lilypond/issues/detail?id=2858
> >> and i'm confused. Do we want to change
> >> \shape GrobName #'offsets
> >> into
> >> \shape #'offsets GrobName
> >> or
> >> \shape #'offsets
> >> (used as a tweak, with GrobName being guessed)?
> >
> > Yes.
> 
> uh, i ususally find that it's a bad idea to answer "yes" to a question
> that goes like "do we want X or Y?"
> :)

The question do we want to change X into Y or Z, and the answer to that was
"yes" because you can now _either_ write Y or Z, and _either_ will work.  \shape
will figure out from the _type_ of the last argument whether it is supposed to
override or tweak.

And since tweaks stack awfully when the tweaked argument is not last, we put the
tweaked argument last in _either_ case.  Even if the tweaked argument happens to
be a grob name (resulting in an override) instead of some music (resulting in a
tweak).

> > It is quite valuable to be able to use this function as a tweak
> 
> ...but from this sentence i deduce that we're trying to do the second
> thing, i.e. \shape will have one #'offsets argument and no argument
> specifying grob name, which will be deduced from the music that
> follows the command.  Right?

Either will work.  That's the nice thing.

> In that case, LGTM.

And in this case?

[janek, 2012-10-03 17:44:42]

On Wed, Oct 3, 2012 at 7:36 PM,  <dak@gnu.org> wrote:
>
>> ...but from this sentence i deduce that we're trying to do the second
>> thing, i.e. \shape will have one #'offsets argument and no argument
>> specifying grob name, which will be deduced from the music that
>> follows the command.  Right?
>
> Either will work.  That's the nice thing.

Wow!  I'm definitely underestimating Lily possibilities here!  Looks
like another improvement of yours that i overlooked ;)

>> In that case, LGTM.
>
> And in this case?

in this case i can only say
_    _  ___ _    _
|   / _  |  |\  /|
|_, \_|  |  | \/ |

Janek