Re: adding to the LSR

[Jonathan Kulp]

Carl D. Sorensen wrote:
> > >
> > Thanks for the offer, Chip.  I've just finished a preliminary run
> > through all of the snippets.  I downloaded the tarball of the entire
> > repo and ran them through the convert-ly script, then did a looping
> > script that ran lilypond on each updated snippet using 2.12.2 and saved
> > the terminal output for each in a text file.
>
> Cool -- I didn't know you were up enough on scripting to do that kind of
> work!  I think such a script should be added to scripts/auxiliar so that we
> have it to use the next time we want to do an update.  (And you just
> demonstrated that you have two of Larry Wall's attributes of a good
> programmer: laziness and hubris.  Good for you!)

This was a very simple script.  I learned a good bit of scripting with 
the help of Patrick Horgan a while back when I was writing my lily2image 
script.  Here's the script for any interested folks (first I did 
"convert-ly -e *.ly" on the whole directory):

#!/bin/bash

#*****************************************#
# Run Lilypond on a lot of files and save #
# the terminal output in text files       #
#*****************************************#

for LILYFILE in *.ly
do
  STEM=$(basename "$LILYFILE" .ly)
  echo "running $LILYFILE..."
  lilypond --format=png "$LILYFILE" >& "$STEM".txt
  rm "$STEM".ps
done

---------------------------

I chose png format b/c the ristretto image viewer on xubuntu allows for 
very quick paging through all of the images in a directory, much quicker 
than going through a bunch of pdfs.

After running the script on the directory with all the snippets in it, I 
run this command to find snippets that didn't compile:

  grep failed *.txt

It reads the terminal output from all the files and brings back the ones 
 that failed.

> > The results were pretty good, really.  About 95% of the snippets
> > compiled and look the same as they did in 2.10.  Two snippets didn't
> > compile at all, but I've already taken care of one of them
> > (adding-octaves-automatically) because I've had to fix that one on some
> > of my own files before.  The convert-ly script took the \octaves command
> > as if it were an octave check instead of a user-defined macro.  I've
> > changed the \octaves command in the original LSR code to \makeOctaves,
> > which will survive the convert-ly update intact.
>
> This probably also indicates a need to change the convert-ly rule for
> \octave.  If it doesn't work for \octaves, it also wouldn't work for
> \octaveAdjustFunction, or some other user-defined variable that starts with
> \octave.  This should be either fixed or added to the issue tracker to get
> fixed.

Yes, I was thinking the same thing but I don't know how to change the 
convert-ly rules.  It was easier for me just to change \octaves to 
\makeOctaves.

> > If you'd like to have a go with a few, try fixing the
> > "changing-the-font-to-smallcaps", which doesn't even work in 2.10 in the
> > snippet currently in LSR.  The font doesn't change at all.
> >
> > Another one that doesn't work in 2.12 is
> > "polymetric-section--synchronizing..."
> >
> > It has a bunch of \InnerStaffGroup commands that must be obsolete
> > because they're causing errors.  Changing these to \StaffGroup makes it
> > compile, but then the staffgroup loses the bracket and it doesn't look
> > right.  If you'd like to fiddle with that code until it works properly
> > in 2.12 that'd be great.  (the authors of the snippet are given as John
> > Mandereau, Reinhold Kainhofer, but these guys usually have bigger fish
> > to fry than snippets in the LSR :) )
> >
> > Others:
> >
> > "setting-control-points-of-slur-manually" compiles but the slur runs
> > right through some of the notes.  This doesn't happen in 2.10 but it
> > does in 2.12.
> >
> > "broken-crescendo-hairpin"--compiles but the hairpin doesn't break anymore
> >
> > "positioning-segno-and-coda-without-line-break"--compiles but objects
> > are colliding
> >
> > There are probably more but I have to do some of my official work now. :)
>
> There are also some snippets that work but should be changed to reflect
> changes in 2.12.  I know of two snippets that have to do with lead sheets
> (Chords, Fret Diagrams, melody, and lyrics) that should be changed to use
> \predefinedFretboards and a FretBoards context, instead of using \markup
> \fret-diagram.
>
> I'm not sure if there are others.  There may be one related to auto-beaming.
> I'm not sure what a good process on this is, but I think at a minimum, we
> should look at NEWS for 2.12 and see if there are any NEWS items related to
> each of the snippets.
>
> That's why I estimated 15 minutes per snippet (I wasn't thinking of
> automation, which I should have).  We should check each of the snippets and
> see if the snippet is made obsolete or should be changed to reflect new
> features added, not just check to see if the syntax is right.

Yes, to do it properly we'll need to examine each one more carefully. 
My idea with this is simply to find out which snippets are broken in 
2.12 and get them at least to compile, so that the LSR could be switched 
 to 2.12 and then the snippets can be checked out more carefully after 
that.  Is that a good strategy?

Thanks for the input, Carl. :)

Jon
--
Jonathan Kulp
http://www.jonathankulp.com