Re: LilyPond 2.22.0 exits with an error if DEBUG option is not given

[Paul Scott]

On 1/31/21 2:52 AM, Jonas Hahnfeld wrote:
> Am Sonntag, dem 31.01.2021 um 02:49 -0700 schrieb Paul Scott:
> > On 1/31/21 1:45 AM, Jonas Hahnfeld wrote:
> > > Am Sonntag, dem 31.01.2021 um 17:34 +0900 schrieb 田村淳:
> > > > Hello Jonas,
> > > >
> > > > > 2021/01/31 17:20、Jonas Hahnfeld <hahnjo@hahnjo.de>のメール:
> > > > >
> > > > > Am Sonntag, dem 31.01.2021 um 16:10 +0900 schrieb 田村淳:
> > > > > > Dear List.
> > > > > >
> > > > > > LilyPond 2.22.0 exits with error for one of my projects.
> > > > > >
> > > > > >
> > > > > > Starting lilypond 2.22.0 [RV_381+BWV_980_score.ly]...
> > > > > > Processing `/Users/tamura/Documents/LilyPond/Vivaldi RV 381 +
> > > > > > Bach BWV 980/RV_381+BWV_980_score.ly'
> > > > > > Parsing...
> > > > > > Interpreting music...
> > > > > > Exited with return code 11.
> > > > > >
> > > > > >
> > > > > > In order to obtain more information, I used —log=DEBUG (in
> > > > > > Frescobaldi “Engrave (custom)…”. Then the project compiled fine
> > > > > > with no error.
> > > > > Hm, very weird. Are you sure that both compilations used the very
> > > > > same
> > > > > path, ie maybe the working one uses a wrapper script that sets some
> > > > > 'magic' environment variables needed to find the right libraries?
> > > > >
> > > > > Also, can you reproduce this on the command line, outside of
> > > > > Frescobaldi? If this happens also in a terminal, the next step
> > > > > would be
> > > > > to get a stacktrace so that we can see where it's crashing. Are you
> > > > > familiar with a debugger to get such information?
> > > > >
> > > > > Jonas
> > > > It is quite weird. I’ve added a bit more notes to the project and now
> > > > the problem is gone. I can’t reproduce the error anymore.
> > > Was it reproducible before the added notes? Might not happen every
> > > time, could be 1 compilation out of 10 or so if it's related to garbage
> > > collection. Can you rewind to the input file that crashed?
> > >
> > > > I haven’t tried outside of Frescobaldi. I used Frescobaldi’s “Engrave
> > > > (Custom)…” with nothing in the “Additional Command Line Options;” vs.
> > > > “—log=DEBUG” in it.
> > > Okay, so no difference in wrapper scripts then...
> > >
> > > Jonas
> >
> > I've had the same problem from time to time.  It's always been
> > reproducible but goes away when I add something.
> >
> > I always use the latest development version (or the latest stable
> > version for those brief periods when it is the latest version).
> >
> > I don't use any wrapper software.  I just edit with Emacs and run from a
> > terminal (using Make most of the time).
> >
> > AFAICT a MWE never fails.
> If you have something that you're able to share (privately if needed),
> I'd be happy to take a look...
>
> Jonas


I believe the problem has always been a segfault.  Since adding 
something has always solved the problem I don't have anything to show.

Paul


Since adding something has