[Priorities]: Issue 884 is a regression

[Alexander Kobel]

[ Sorry, attachments were to large in the first attempt - see the link 
below. ]

Dear bug squad,


following the discussion "spacing/breaking issue" on -user, I'm sure 
that issue 884 <http://code.google.com/p/lilypond/issues/detail?id=884> 
is a regression, but not classified as such.
For verification, I [attached] created output from 2.12.3 (Ubuntu Lucid 
stable) and 2.13.22 for the very snippet in the bug tracker; 2.12.3 
renders two pages as expected, 2.13.22 makes it four.  Check out the 
files under 
<http://lists.gnu.org/archive/html/lilypond-user/2010-06/msg00001.html>; 
they're just too big to post them on -bug.

IMHO, the earlier variant is the expected one, so it should not be a 
documentation issue as mentioned in the bug report:  Even _if_ you 
wanted to specify a certain number of pages between \pageBreaks, you 
often don't want the _same_ number between all manual breaks (which is 
what you actually specify via page-count as of 2.13.x).  Rather, you 
want to add \pageBreak at "critical" points like the beginning of a new 
section or a really good page turn for performances, and leave the rest 
up to Lilypond.
Besides, page-count just implies that you specify the total number of 
pages, doesn't it?

As an additional enhancement request, it'd be nice to have a "localized" 
variant of page-count between \pageBreaks.  E.g., I could imagine to 
allow the definition of page-count either as an integer that gives the 
total number of pages (expected behaviour until 2.12.x), or as a list of 
integers specifying the number of pages between consecutive \pageBreaks. 
 If \pageBreak really divides the piece into different chunks, as Chris 
suggested in the bug report, the latter might actually be easier to 
implement than the first variant.


Cheers,
Alexander