Menu
▾
▴
#1173 Segfault with showFirstLength/showLastLength when skipBars is set
Verified
nobody
None
Defect
2010-10-26
2010-07-05
Anonymous
No
Originally created by: *anonymous
Originally created by: brownian.box@gmail.com
Reported by James Lowe:
http://lists.gnu.org/archive/html/bug-lilypond/2010-07/msg00012.html
(there is some discussion too)
%-----------8<-------------------------
\version "2.13.26" % 2.12.3 crashes too
MusicOne = {
\time 3/4
\compressFullBarRests
[r2].*11 \bar "||"
}
showLastLength = [r1]*2
\score {
\new Staff {
\MusicOne
}
}
%-----------8<-------------------------
This gives:
%-----------8<-------------------------
$ LANG=C lilypond sample.ly
GNU LilyPond 2.13.26
Processing `./sample.ly'
Parsing...
Interpreting music...
Preprocessing graphical objects...
Finding the ideal number of pages...
Fitting music on 1 page...
Drawing systems...
././sample.ly:6:4: programming error: Spanner `MultiMeasureRest' is not fully contained in parent spanner. Ignoring orphaned part
[r2].*11 \bar "||"
Segmentation fault
Originally posted by: n.putt...@gmail.com
The presence of \score is unimportant here.
showLastLength = [r1]
{
\time 3/4
\compressFullBarRests
[r2].*2
}
Summary: LilyPond crash when using showLastLength with \compressFullBarRests
Originally posted by: pnorcks@gmail.com
Also note that no segfault occurs when the multi-measure rest "length" agrees with the time signature:
showLastLength = [r2].
{
\time 3/4
\compressFullBarRests
[r2].*2
}
Originally posted by: pnorcks@gmail.com
That was probably an obvious comment. I don't really use showLastLength that often... :)
Originally posted by: pkx1...@hotmail.com
Just to re-itterate from the email thread.
In cases where you use ShowLastLength for a score that has multiple scores in it and that all have different time signatures then this can be a problem.
What I found specifically (to give this context) was that if I used [r2]. then it didn't crash on ANY time sig that was not necessairly matching (such as 4/4 or 2/2) I got odd output but no unexpected errors, but if I used [r1] it always crashed when one of the scores was 3/4. I hadn't experimented with any other time sigs
Originally posted by: n.putt...@gmail.com
Mike Solomon has started work on this: http://codereview.appspot.com/1741060/show
Labels: Patch
Status: Started
Originally posted by: n.putt...@gmail.com
(No comment was entered for this change.)
Summary: LilyPond crash when using showFirstLength/showLastLength with \compressFullBarRests
Originally posted by: n.putt...@gmail.com
Mike's latest patch: http://codereview.appspot.com/1866051
Originally posted by: Carl.D.S...@gmail.com
So what is the status of this fix? At least part of the latest patch appears to have been applied to origin.
Neil sent out a rebased patch in a mail to a few of us (Mike, Graham, me). Is that patch considered a push candidate?
Should a new issue be raised on Rietveld with Neil's latest patch?
This is the *last* critical issue remaining (except the web, of course, which won't be fixed until we go to 2.14)
Originally posted by: percival.music.ca@gmail.com
slight correction: I'm going ahead with the alpha test now, and then I'm switching web. So rather: this is the last critical issue remaining before we can have a release candidate and begin the two-week waiting period.
I'll send an explanation of the current release process plan to -devel once .34 is uploaded; I will accept objections and possibly modify the plans.
Originally posted by: v.villenave
[Oops. I sent my comment before it was complete.]
I've been trying to review Mike's patch; it looks pretty good to me but as Neil pointed out is does come with a cost in terms of calculation, and solving this specific issue is out of my depth.
However, one simple workaround (however inelegant) would simply to *unset* skipBars whenever show[First|Last]Length is set. Which I'd be happy to do; afterwards it would just come down to opening a new issue entitled "Enhancement: make showFirstLength compatible with skipBars". Unless we hear from someone with decent coding abilities soon, I think getting rid of the segfault is the priority here.
Originally posted by: percival.music.ca@gmail.com
The historical fix for segfaults is simply to output an error message ("you can't using showFirstLastLength and skipbars") and then quit cleanly. I'm fine with unsetting skipBars -- with a warning message -- as well.
I definitely don't think that "making showFirstLengh work well with skipBars" is a Critical issue.
Originally posted by: v.villenave
OK, then I'll get on it and open a new issue when I'm done.
Actually, it's a bit more complicated than I thought, since skipBars may be set later in the .ly code (e.g. if for some reason the user sets \compressFullBarRests halfway through their piece).
I've cleaned the code a bit (which does break Mike's patch, and I apologize for that even though I'm not sure this was the way to go anyway). I have a couple other ideas on how to deal with this problem, even though some of these may be slightly beyond my reach :-)
Owner: v.villenave
Originally posted by: v.villenave
The problem is actually much more serious than it looked, and Mike's patch actually doesn't address the underlying cause.
I'm opening a new issue 1336 to address the underlying cause of the problem. Anyway, it's now officially out of my league.
Mergedinto: 1336
Labels: -Patch
Owner: ---
Status: Duplicate
Originally posted by: v.villenave
(No comment was entered for this change.)
Mergedinto: -1336
Status: Verified