Re: Understanding Lilypond

[Urs Liska]

Am 17.01.2015 um 13:09 schrieb Peter Gentry:
> Thanks for your reply Kieran - I'm not sure that my response is suitable for 
> the list but will post it there anyway. Please feel
> free to critisize scorn or otherwise flame.

Yes, this is suitable for the list.

> 1. This is difficult. My background is chiefly in Fortran and Visual Basic 
> although I have dabbled with machine code. These codes
> have well defined structures variables are defined there are multi dimensional 
> arrays and subroutines are similarly well defined as
> far as parameters and scopes are concerned.
>
> When I look through the Lilypond source code I see
>
> (a) at least three different languages being used C++, Scheme, and Python. It 
> is not clear how these are integrated or why they are
> needed.

C++ and Scheme are the languages that are used for LilyPond itself.
I can't fully comment on that but at least I can tell you that Scheme is 
a fundamental part of GUILE, the GNU framework or application environment.
One advantage of using Scheme as opposed to a plain C++ program is that 
very much can be done by users, without needing to recompile the 
program. The more of LilyPond's own working can be done with Scheme the 
more fundamental extensions can be created in the user space.

If I'm not completely mistaken Python is only used for separate scripts 
like convert-ly or lilypond-book, but also for the build system.

> (b) the syntax and scope of variables is hard to follow through the various 
> procedures probably mainly due to a lack of
> understanding of the way scheme objects are stored in memory.
> I have started to learn these codes but often find various definitions and 
> example scripts only work in one variety of the code.
>
> (c) As with most learning tasks it is important to start out with correct 
> descriptions and definitions. The tyro can be led astray
> by ill founded assumptions as to how the code elements work and fit together. 
> In short you can see difficulties that don't exist or
> try to implement unnecessary coding. A simple example being lists and 
> regression in scheme the use and philosophy is different from
> say Fortran and similar languages where the basic principles allow you to 
> "guess" how to use unfamiliar procedures rather as you
> would use paper and pencil. You can often read code and see at a glance the 
> gist of what is going on.
>   (d) summing up when looking at the source code there is a dazzling array of 
> procedures but how and when they are called is not
> clear - nor how they relate to each other. I would be looking for a main 
> programme and branches to various sub programmes or
> routines I.E. a flow diagram!

In short I'd say you're still struggling with the fundamental principles 
of Scheme. Which is more than natural - I think the number of people 
having mastered this is only a small fraction here (and I definitely 
don't counte me to that fraction).
So the question would boil down to the question: why Scheme?
As said above it's the official extension language of the framework, so 
there's not much of a choice here.
But I start to comprehend that Scheme is quite "generic". This is what 
makes it so hard to learn but this also makes it extremely versatile and 
fundamental.

> 2.
> (a) The philosophy is outlined but is short on detailed explanation in terms of 
> code. The ideas of engravers, grobs etc is
> understood but the details of their parameters and how they are altered in ly 
> scripts is hard to find in a systematic way.
>
> There are excellent aids to get one writing ly scripts and producing quite 
> complex scores but not to make those individual tweaks.
> Looking at the various tweaks published on the list can be bewildering the 
> common first impression is why? how? Where was that in
> the documentation?
>
> Maybe I'm just to long in the tooth for grasping new concepts. More experienced 
> programmers will rightly say we are not spoon
> feeding and you should put in the hard graft yourself. I agree with that but 
> would love some signposts to reduce the feeling of
> being lost in a sea of hieroglyphs.

This is exactly what we have been discussing about more than once (and 
actually currently in a private discussion).
I fully agree with you that there should be much more documentation 
material giving users help with the challenge to walk over that step. 
The problem is simply that there are too few people who are able to 
create that kind of documentaion, and those who are there are probably 
too busy doing other work.

Concretely I see the problem in a sequence of related issues:

- Scheme itself *is* difficult to get into
- There are so many Scheme dialects, and it is confusing to find out 
which information would actually apply to the Scheme in LilyPond
- The way to interact with LilyPond through Scheme is quite obscure
  (and that's where better documentation would be needed most IMO)

Urs

> > -----Original Message-----
> > From: Kieren MacMillan [mailto:address@hidden
> > Sent: Friday, January 16, 2015 9:40 PM
> > To: Peter Gentry
> > Cc: Lilypond-User Mailing List
> > Subject: Re: Understanding Lilypond
> >
> > Hi Peter,
> >
> > > many of us have struggled for many months to get to grips
> > with the structure and philosophy of Lilypond.
> >
> > 1. Regarding the structure, what are you struggling with exactly?
> >
> > 2. Regarding the philosophy, what are you struggling with exactly?
> >
> > Hope I can help!
> > Kieren.
>
> _______________________________________________
> lilypond-user mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/lilypond-user