Sat 02 Feb 2013 11:12:22 PM UTC, original submission:
The sequence
leads to
The above is with a debugging build of octave 3.6.4-rc1.
I can reproduce this with an optimised build on the same machine and octave 3.6.4-rc0.
I can also reproduce this with an optimised build on another machine and octave 3.6.3.
I have another test case, though, i.e. the matrix X from data_X, which leads to a segfault on two of those three setups but not on the optimised 3.6.4-rc0.
When it does, the crash only takes a second or two to occur (with larger matrices, it can take quite a while, so this is already quite the reduction.
This is on Exherbo Linux with a 3.7.2 kernel.
The other machine I mentioned above is Debian with a 3.2.0-0.bpo.1-amd64 kernel.
Data's at
http://page.mi.fu-berlin.de/pipping/data_X
http://page.mi.fu-berlin.de/pipping/data_N
(because savannah always gives me an error when I try to upload it)
|