Re: Segfault 2.13.47

[Jay Anderson]

On Sun, Jan 30, 2011 at 11:27 AM, Keith OHara <address@hidden> wrote:
> I wish you luck in making a small example.
>
> Maybe you can test an earlier Lilypond version on your complete score?
>
> Do you use Reinhold's new (not yet documented) \partcombineTogether, etc.,
> that appeared in 2.13.35 ?
>
> To me, the stack trace hints that something went wrong in
> Part_combine_iterator::kill_mmrest(), or data it depends on.  The only
> recent changes to that piece of code were cosmetic, and appeared in 2.13.39
> and 2.13.41.

It actually occurred somewhere between 2.13.45 and 2.13.46 from what I can tell.

More info: I have two parts in the score like this: \partcombine
\bassoonOne \bassoonTwo where not all of \bassoonTwo is filled in
(only a few measures at the beginning are there). But isolating the
score down to just these two parts doesn't result in a segfault. Even
taking out an unrelated part from the score makes the segfault go
away. When I filled in the rest of the part the segfault went away so
this bug may not be likely to happen under normal circumstances (where
both part in partcombine are the same length).

I've attached the whole set of files which cause this since I've been
unable to make a small example. 'score.ly' causes the segfault. Seeing
the segfault was somewhat disconcerting, but it may not especially
severe since it is difficult to isolate and it went away when
completing all the parts.

-----Jay

GlazunovReveries.tar.bz2
Description: BZip2 compressed data