Re: FW: Professional ???

[Jan Nieuwenhuizen]

"Jeff Henrikson" <address@hidden> writes:

Hi Jeff,

> I didn't cc it to the list because it was somewhat composition
> specific and was extremely critical, but he said he found it useful,
> and as I read it again, I think the criticisms are harsh but fair.

As you observed yourself, current lilypond input is not well suited
for composing and we (seem to) have a different whishlist than a
contemporary composer might have.

However, I think this has more to do with focus and pragmatism, rather
than philosophy: we want to do well at typesetting, so we concentrate
on that.

Also, the problem of composing on the computer hasn't been solved.
Maybe GUIs can bring a solution, some day.  You tried using a
spreadsheet (you should have used Gnumeric :-).  Maybe a smart emacs
mode could help.

So, let's try to be pragmatic.  Criticism is fine, but doesn't really
help if it's too vague. 

> - lily handles conventional notation reasonably, but degrades very
> quickly as soon as esoteric needs arise.  Don't ever try to enter a
> Boulez or Stockhausen score without expecting to need psychiatric
> care!

Probably.  We've heard this before, but it doesn't help.  You should
try to request a specific feature or notation that you really need.

> - the concept of embedding a two dimensional entitiy like a musical
> score in a one dimensional text file seems suspicious from the
> start.  It requires huge amounts of effort to get reasonable
> results.  I personally don't understand why they don't read their
> source code out of text in a spreadsheet.

For Lilypond this won't make a difference.  A spreadsheet is just as
1-dimensional as a .ly file.  You proved that yourself, in creating a
.ly from a spreadsheet.  So, maybe a smart editor would help the user,
and you're encouraged to investigate that.  But it has no effect on
the typesetting (capabilities), so *we* will probably not look into
this.

> the processing time of the attached score is over 10 minutes with
> build 1.3.46 on my 266 MMX pentium.  The newer builds are much
> slower.

Some improvements were made, but Lilypond is a memory hog.  If you're
out of ram when processing this score, adding some would really speed
up the processing.

> This is very frustrating when a rhythmic error has been made because
> rhythmic errors generally freak lilypond out

This can be frustrating.  I used to check each part of the coriolan
separately, before generating the score.  Also, we now have an emacs
mode that help you correcting lilypond errors.

> - entering alphanumeric text around the score appears to be very
> limited, though I do not claim to be a super-user.  For example, (to
> my knowledge) you cannot have two different typefaces, one for tempo
> and one for a style marking, and get them to coexist vertically.

This has been addressed in recent releases by allowing scheme markups
in text.

> - percussion support is basically nonexistent.  Add to that the fact
> that percussion parts have lots of text on them, sometimes with
> conventions dictating weird locations like below staves.

What exactly do you need for percussion?

> - extracting parts is a very laborious process.

Have you looked at the coriolan example?  If you set everything up in
an orderly way, generating parts and score should be fairly easy.

> - For undetermined reason, I have not been able to get multimeasure
> rests outside of a trivial example, critical for extraction of
> individual parts.  I seem to be following the documentation/examples
> but the behavior is not consistent and 10 minute compiles are not
> conducive to experimentation.

You are aware of the skipBars property?  See
input/test/multi-measure-rest.ly.

> - changing the paper/font size is horribly braindead.

This should have been fixed in later releases.

> Landscape rotation likewise.

Ly2dvi has got a --landscape option.  You'll probably have to set the
linewidth by hand.  This should not be hard.

> - (jazz specific) the chord symbol system is totally useless for
> sophisticated chord symbols.

The chord system was rewritten recently.  There's a jazz-chords
example, but I don't know too much about chord schemes.  James Hammons
can maybe give you some pointers.

> - getting the symbol c1. to span two measures of 6/8 time seems to
>   be impossible.

For two measures of 6/8, you probably need to enter something like
c1*12/8

> In summary, it has been an extremely frustrating process learning
> and some of the features I really need I know won't be implemented

Try your luck at http://appel.lilypond.org/wiki?FeatureWishes

> - Barchecks were annoying to insert by hand.

You do now that you don't need bar checks?

> - Lily knows how to expand a R1*n into little R1s, but it doesn't
> know how to congeal multiple R1s into a big R1*n

This would be a valid feature request, I guess.

> with rehearsal marks is absolutely necessary when handing parts to
> human players.

Is there a problem with rehearsal marks?

Greetings,
Jan.

ps: please fix the line length of your postings

-- 
Jan Nieuwenhuizen <address@hidden> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien       | http://www.lilypond.org