Re: LilyPond benchmarks - request for proposals

[Jan Warchoł]

Hi,

i was cc'ed, so despite my "official" absence i'll reply :-)

2014-03-01 12:01 GMT+01:00 Urs Liska <address@hidden>:

Hello LilyPond users,

LilyPond's declared claim is "automated engraving", and we're often talking about how exceptionally good its default results are. But we're also often talking about stuff that isn't good enough by default (see for example Janek's blog post about how many slurs and ties he had to fix for our big Fried edition).
[...]
These benchmarks would automatically be built for any new version of the website and give us a clear picture of what has to be done to achieve perfection.

Today I'd like to ask for suggestions: a) to assemble a set of score _types_ to be used in that benchmark series, b) possibly for concrete suggestions.

A choral piece:
https://github.com/janek-warchol/eja-mater-demonstration

In the first commit (566213f) you can see the "raw" (no manual adjustments) version, and the following commits add adjustments one-by-one, often with rationale in the commit message.  I think this is a very good resource for benchmarking.

Another two examples are here (too big for an attachment, so they'll be availabe in my Dropbox for a couple weeks)
https://www.dropbox.com/sh/6v9z7tapu5n48k6/htaH4MQlcQ
There's one solo violin fragment with many problems marked, and another choral piece with a lot of markings (i know that the pdf is compiled with an old lily version, but >90% of the problems marked there remain in 2.19.3). Unfortunately the sources for the choral piece are partly broken - lyrics don't align with appropriate notes (but it's easy to fix that).  The meaning of the colored markings is explained in another pdf.

I hope it'll be useful.

best,
Janek