[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: "insane spring constant" error
From: |
Valentin Villenave |
Subject: |
Re: "insane spring constant" error |
Date: |
Sun, 29 Jul 2007 02:25:08 +0200 |
2007/7/28, Joe Neeman <address@hidden>:
> That's funny, I can't reproduce it (with or without the
> Time_signature_engraver). Do you have another example that causes the
> problem?
Jean-Charles has encountered problems too when compiling the
Documentation (he says almost every snippet produced this message).
I was running LilyPond 2.11.28 on linux x86 (now I'm back to 2.11.27,
but I'm reinstalling .28 to see if I can reproduce it).
Nearly every score I tried to compile produced this message, in fact.
And each time, I could "fix" it by removing something: the
Time-signature_engraver, the SlashSeparator thing, or the tremolo
repeats (see my previous mails)...
OK, I've just reinstalled .28, and it's doing it again with both
samples I posted yesterday. "programming error: insane spring
constant" (the output is fine though).
Weird.
In addition to the two snippets I already posted, I tried to launch a
benchmark with several scores I found on Mutopia, using different
versions of lilypond, on different OS (32 and 64 bit linux versions)
It seems to happen exclusively with "large" scores (i.e. orchestral scores)
2.11.28: the "insane" message was displayed about 600 times, with a
huge huge huge CPU activity (the first try, with another score, made
my computer crash after 50 minutes) -- I have 1024 Mo of RAM, and no
swap partition; maybe this is a possible explanation?
2.11.27: no error message, the compilation went right and I was still
able to use firefox when it was working. My best results were with
2.11.27 64bits version; it's really amazing!
2.10.25: a few minor errors due to obsolete code, but it was fine (not
as fast as 2.11.27 though)
Besides, I noticed compilation went much faster with .27 than .28;
e.g. 2 minutes vs 3 minutes for .28!
OK, now I've just spent a couple hours testing, installing,
uninstalling different versions at least 30 times, so I hope we'll
find why this is happening...
Obviously, something was changed in .28 code -- at least, now I
understand what "development version" means ;)
Regards,
Valentin