lilypond-user
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Strange LilyPond crash


From: Thomas Morley
Subject: Re: Strange LilyPond crash
Date: Tue, 24 Nov 2015 00:12:28 +0100

2015-11-22 1:59 GMT+01:00 Thomas Morley <address@hidden>:
> 2015-11-21 17:31 GMT+01:00 Thomas Morley <address@hidden>:
>> 2015-11-21 16:42 GMT+01:00 David Kastrup <address@hidden>:
>>> Marc Hohl <address@hidden> writes:
>>>
>>>> As soon as I remove an instrumental part at random (or even some
>>>> notes), the crash may or may not disappear. I am not sure but IIRC a
>>>> file that compiled fine during one run crashed without having done any
>>>> changes to that specific file (nor its includes).  Is it possible that
>>>> the self-compiled version I am using has some internal errors that may
>>>> cause the crash? I update the git repository regularly and re-compile
>>>> lilypond after applying the latest patches.
>>>
>>> Both 2.19.29 and 2.19.30 had really bad problems that could be
>>> exacerbated by "large files".  Your segfault does not look related to
>>> that as far as I can tell.
>>>
>>> --
>>> David Kastrup
>>
>>
>>
>> Hi Marc,
>>
>> I'd like to offer you send me the zipped files offlist.
>> Maybe a second pair of eyes watching the real code may help.
>>
>> Cheers,
>>   Harm
>
> Hi Marc,
>
> thanks for your files.
> Up to know I can't come up with a tiny example.
>
> Though, some observations:
> (1)
> All works with 2.18.2 (after downgrading the syntax)
> (2)
> Replacing all occurances of
> \repeat percent ...
> with
> \repeat unfold ...
> make it work with a build from latest master
>
> Thus I suspected
>
> commit [rb416f10429d8d3881445d9000ff422dc67176df1]
> Author: David Kastrup <address@hidden>
> Date:   Wed Jul 15 23:30:30 2015 +0200
>
>     Issue 3693: Let Percent_repeat_iterator be unfazed by Timing changes
>
>     There is still one shortcoming: the percent repeats will not contain any
>     material apart from the percent itself.  In particular, no Timing
>     changes will be repeated.  If there are meter changes or \partial
>     commands inside of percent repeats, they need to occur in parallel
>     passages outside of any percent repeat, if necessary in a separate
>     "timing track".
>
> Which was the last commit I found for this area of code.
>
> Though a build from one commit before:
> cd3c64ba5131d64d7f0b76487054f8fac1364e61
> returned the same segfaults.
>
> Will continue testings tomorrow.
>
> Cheers,
>   Harm

Ok, made a mistake.

One commit before is:
11a9e5701316d46d34993fadb85482282c9753cc

A build from this one works!

So,
commit b416f10429d8d3881445d9000ff422dc67176df1
Author: David Kastrup <address@hidden>
Date:   Wed Jul 15 23:30:30 2015 +0200

    Issue 3693: Let Percent_repeat_iterator be unfazed by Timing changes

    There is still one shortcoming: the percent repeats will not contain any
    material apart from the percent itself.  In particular, no Timing
    changes will be repeated.  If there are meter changes or \partial
    commands inside of percent repeats, they need to occur in parallel
    passages outside of any percent repeat, if necessary in a separate
    "timing track".

is indeed the problem.


Now it would be nice I'd could come up with a tiny example...

Though, very strange things happen, like:

assume file-1.ly with

foo = { some-music }
buzz = { some-other-music }

and

file-2.ly with

\include "file-1.ly"
\score {
  <<
  \new Staff \foo
  \new Staff \buzz
  >>

compiling file-2.ly causes a segfault.
commenting \new Staff \buzz still causes a segfault
additional commenting(!!!!!) buzz = { some-other-music } in file-1.ly,
and the segfault disappears!!!
But it's not consistent! It depends what else is defined in those files.

No idea how to boil it down to a small example.


Sorry for the initial wrong info,
  Harm



reply via email to

[Prev in Thread] Current Thread [Next in Thread]