Re: The composition process doesn't get disrupted using Lily Pond?

[H. S. Teoh]

On Wed, Jan 09, 2019 at 06:06:27PM -0800, Flaming Hakama by Elaine wrote:
[...]
>    So, the tl;dr version is that, yes there can be blocks to the
>    composition process when using lilypond.  But it is often possible
>    to bucket that work, so you can get it done and get back to the
>    composition process.  I think it can be worth it since there are a
>    lot of good features in lilypond to help composition/arranging.
>
>    My composition process sometimes includes lilypond.  Generally
>    speaking, this is only at the point when I know the entire layout
>    of the piece--at least in terms of the overall structure, the
>    sections, and their durations.
[...]

Yes, my workflow when composing using lilypond is similar to this.


[...]
>    The reason I like to start with the complete roadmap planned out is
>    that it can be a pain to update the durations.  Especially if there
>    are many instruments in the scores. 

My approach is to use unique section IDs as comments in \global, e.g.:

        global = {
                \tempo ... % blah blah

                % Intro
                s1*12

                % Main theme
                s1*24

                % Episode 1
                s1*18

                % Main theme (2)
                s1*20

                % Episode 2
                s1*22

                % Conclusion
                s1*9
        }

and then repeat these comments verbatim in each instrument's part.
These comments become the milestones from which I can count measures
when adding/deleting measures, to make sure things line up (and add up).
I generally also group my measures into blocks, generally 4 measures per
block, to facilitate quick counting.  These marker comments also make it
easier to find the right line when I need to edit something.


>    For example, if you then want to add a measure to one section of
>    the song, you need to add this in the global definitions, as well
>    as in each instrument/staff's music variable (and if you have chord
>    symbols, it will need to be updated there, too.)   So, there
>    becomes a little more overhead to modifying the duration of a
>    section after the fact.  This type of thing can be done, but it
>    does feel like a break in the composition process since you go from
>    musical entry to lots of cutting & pasting in multiple files, with
>    all of the debugging to ensure that you did it correctly.

My personal preference is to use a single file for all parts.  Yes, I
know that sounds crazy, but I find that it really helps in keeping
things in sync, rather than scattered across multiple files that are
even harder to keep track of.  Using a text editor with nice search /
bookmarking / split-window features makes this a lot more usable than it
sounds (if you ever have to reach for the scrollbar, you're doing it
wrong).

Using my landmark comments approach, it's a simple matter of searching
for the landmark string to locate the appropriate section.  And you
could also prefix / suffix each landmark with the instrument
abbreviation, say "Episode 2/Fl", "Episode 2/Vln1", then to you could
jump to the landmark for Episode 2 for the 1st violins with a single
search, to locate the measure that needs editing.  This, plus copious
use of bookmarks, let you jump between different instruments of the same
passage easily, which reduces a lot of the tedium in editing your score.


>    On the other hand, you can certainly use lilypond to sketch out
>    ideas, not worrying about, for example, how many bars will be in
>    the intro while you work on the main theme, or how many instruments
>    will be in the ensemble while you're trying to figure out the
>    harmonic progression.  But at some point you will need to stop
>    working on a sketch, and proceed with engraving.  At which point,
>    you will have another break in the composition process while you
>    prepare your lilypond code for note entry.

IME, I find that it works better if you do your lilypond sketching as
something completely separate from the actual, final score.  Use it to
explore ideas, get a feel for duration, section lengths, etc., in sketch
form (short score, piano reduction, or even just a single running staff
with tentative notes).

But once you're done with sketching, put it aside and begin a new
lilypond score from a clean slate, lay out your instruments, etc., then
copy-n-paste the music from the sketch into the various blanks.  I find
that this is far less messy than trying to hammer a rough sketch score
into something presentable.  It also liberates you in the sketching
phase from worrying about final score layout; you can sketch out new
ideas and experiment with throwaway ideas, hack the score without regard
to "proper" lilypond score practices, without fearing you'll ruin the
structure of your final score.


>    So, generally, I would say that I tend to use lilypond after most
>    of the composition has been done.  But you can use it earlier in
>    the process.  It's just that it is likely you will have to refactor
>    your sketches into a final engraving at some point, which will be a
>    break from the composition process.
[...]

I've done both, and I've found that I generally would compose the rough,
over-arching structures first, before getting into lilypond, but the
workflow I've setup works well enough that pretty early on in the
composition process I'm already inputting things into lilypond.  Editing
is more tedious, for sure, but being able to see in-score what you have
and having the option to use the midi output for proof-reading helps a
lot.

One thing I find that helps, is to do complicated editing as a separate
lilypond project rather than trying to do surgical work on a full-scale
lilypond score.  E.g., if I decide somewhere along the line that I don't
like a particular passage and want to replace it, it helps to work out
the replacement passage in a separate sketch score first.  Esp. for
orchestral scores where adding/removing measures can mean editing 16-20
staves simultaneously, it really helps to work out all the details of
the replacement passage, most pertinently the exact duration, first, and
then come back to make exactly that much space in the master score (and
make sure everything all lines up) before pasting in the actual notes
from the sketch score.

Also, I highly, *highly* recommend using version control to manage your
lilypond score (if you aren't already).  Manual backups are prone to be
lost / accidentally deleted or edited, etc., and are a pain to manage.
Lilypond's textual input format is especially well-suited to revision
control systems like Git (which I use).  Using a VC system lets you
boldly experiment with new ideas without fearing that you'll somehow
mess up / lose your current work. It gives you the full history of your
work, which you can revisit at any time e.g. if you remembered an
earlier idea that you discarded but later on decide was a good idea
after all. You can even use different branches to manage different
editions of your score (e.g., slightly different instrumentations,
arrangements, etc.), and use the VC's merging features for syncing
between different editions.  I wouldn't dare compose in lilypond without
using version control!


T

-- 
Famous last words: I wonder what will happen if I do *this*...