lilypond-devel
[Top][All Lists]
Advanced

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

Re: Even more disconnected


From: Han-Wen Nienhuys
Subject: Re: Even more disconnected
Date: Sat, 19 Aug 2006 19:29:14 +0200
User-agent: Thunderbird 1.5.0.5 (X11/20060803)

Erik Sandberg wrote:
On Tuesday 08 August 2006 22:01, Han-Wen Nienhuys wrote:

Hi,

Erik Sandberg wrote:
Hi,

I will be completely away from all kinds of computers whatsoever for
about one more week. But stay tuned: some radical cleanups of music
functions are in progress.
did you realize that your last radical cleanup left a memory leak that
brought my autobuild to its knees?
Ouch, sorry!
This type of error is caught by make web, as it runs input/regression/ in one single lilypond invocation.

Strange: my gs problems make lily stop at regression/utf-8.ly, which among the last regression/ files; so I should have noticed the problem here too. Perhaps the problem is that I have too much RAM? (1G)

Perhaps it would be sensible to let `make web' constrain lily's memory usage by default, to something reasonable like 256M?

Yes, good idea.

Can you put a priority to fixing your GS configuration so you can run "make
web" completely, and fix these problems before they are committed?  Thanks.

Is this the scm_gc_protect in translator.cc (which you already fixed), or is it something unknown?


It's unknown. The symptom is that -ddebug-gc shows that over the course of a lot of files, some objects seem to accumulate, in particular Probs and Moments.

(in the former case, I don't understand the error completely: how can the GC protection introduce a memory leak?)

If protect something, but don't have a matching unprotect, then memory will never be released.


I am now back in civilization, so I can work on the gs problems. I'll try starting with a fresh 32-bit ubuntu installation.



--

Han-Wen Nienhuys - address@hidden - http://www.xs4all.nl/~hanwen

LilyPond Software Design
 -- Code for Music Notation
http://www.lilypond-design.com





reply via email to

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