Re: crash from bleeding edge sources

[Leo Butler]

On Wed, 20 Oct 2010, John W. Eaton wrote:

< On 20-Oct-2010, Leo Butler wrote:
< 
< | Hello,
< | In building from the bleeding edge sources this morning, I encountered the 
< | following at the end of the build:
< | 
< | ../../run-octave -f -q -H -p . --eval "sparseimages ('gplot', 'txt');"
< | *** glibc detected ***
< | /home/work/octave/sandbox/octave/src/.libs/lt-octave: free(): invalid
< | pointer: 0x418de900 ***
< | ======= Backtrace: =========
< | /lib/i686/cmov/libc.so.6(+0x6b321)[0x42fa7321]
< | /lib/i686/cmov/libc.so.6(+0x6cb78)[0x42fa8b78]
< | /lib/i686/cmov/libc.so.6(cfree+0x6d)[0x42fabc5d]
< | /usr/lib/libstdc++.so.6(_ZdlPv+0x21)[0x4313f701]
< | 
< | The complete output is viewable at
< | http://www.maths.ed.ac.uk/~lbutler/crash.html#crash
< 
< Did you update from a previous build, or is this a completely fresh
< install from scratch?
< 
< If you updated, it's possible that changes in gnulib are what is
< causing trouble.  The simplest thing is to try a completely new build
< from scratch.  Or, you can also try to wipe out the libgnu directory
< in your build tree, then start with autogen.sh and configure.
 
 John, I tried the solutions in succession. The only one that worked
 was a completely new build.

 I'm curious if you could suggest a way for me to avoid rebuilding
 from scratch.

 Thanks,
 Leo

-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.