You are not logged in.
[~]# matlab
/bin/sh: error while loading shared libraries: libdl.so.2: cannot open shared object file: No such file or directory
[~]#
Does anyone know what this is referencing? I just noticed this error after a pacman -Syu, after about 6 months. I remember why I should only upgrade individual packages one at a time, and only if the new package possesses a new function I desparately need.
This last major arch upgrade seems to be less succinct and efficient as the previous ones for me. I had to basically troubleshoot about 5-8 faulted apps in the last week.
Offline
Dont know what the problem is, but Matlab works fine on my Arch install: Version 7.1.0.183 (R14) Service Pack 3
Offline
pacman -Qo /lib/libdl.so.2
/lib/libdl.so.2 is owned by glibc 2.4-4
HTH
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
(A works at time B) && (time C > time B ) ≠ (A works at time C)
Online
lessthanjake, I am running version 7.0.1... is there a way to upgrade. If I disable LD_ASSUME_KERNEL=2.4.21 in the matlab start script, it works but all symbolic processing doesn't function anymore. What discipline do you use matlab for by the way?
Offline
Is your system up-to-date? You should have that file on your system. It's part of the current glibc package as Lone_Wolf pointed out.
Offline
Yes, my system is up to date. I also do have the file in /lib. I think my problem is that the system is up to date and have some issues with the new glibc package.
Offline
Try running pacman -Syu again... I'm getting the feeling you have an out of sync mirror.
Dusty
Offline
LD_ASSUME_KERNEL to force linuxthreads doesn't work anymore, linuxthreads is dead and forcing a 2.4 kernel to LD will cause it to skip all glibc libraries, resulting in the failed link message you get.
Offline
Dusty, I did pacman -Syu again. But no luck.
JGC, disabling LD_ASSUME_KERNEL allows me to start matlab -- however, many things seem broken like the symbolic package.
in matlab:
>>syms a b
>> a + b
Unable to load mex file: /opt/matlab/toolbox/symbolic/maplemex.mexglx.
/opt/matlab/bin/glnx86/libmaple.so: symbol errno, version GLIBC_2.0 not defined in file libc.so.6 with link time reference
??? Invalid MEX-file '/opt/matlab/toolbox/symbolic/maplemex.mexglx': .
Error in ==> maple at 104
[result,status] = maplemex(statement);
Error in ==> sym.maple at 85
[result,status] = maple(statement);
Error in ==> sym.plus at 22
X = maple(A(:),'+',B(:));
I refered to this before though. I'm still not sure whats the hold up.
Offline
So I'm suffering from the exact same problem as clam. Basically what I've gleaned from the internet is that maple (the backend for matlab's symbolic functions) isn't compatible with nptl threading, only linuxthreads.
According to this thread: http://bbs.archlinux.org/viewtopic.php? … sumekernel
and JGC, arch doesn't support linuxthreads anymore, so....
Does this mean arch cannot support maple/symbolic matlab anymore?
Offline
Maybe it is possible:
Alternatively, obtain a version of the C library that does not have support for 2.0 symbols removed -- glibc-2.2.5 is a good one. Place this library into $MATLAB/sys/os/glnx86 for MATLAB's use, where $MATLAB is the root MATLAB directory.
from:
http://www.mathworks.com/support/soluti … 1BDU5.html
the first possibility they talk about is to use LD_ASSUME_KERNEL that doesn't work anymore, but it is still possible to build the old glibc-2.2.5 from an old PKGBUILD and put it in the $MATLAB directory making it use the old version without the risk to break the whole system.
I am going to try it...
Offline
I had the sample problem while installing maple. I searched the internet and I found this solution. I believe it was also written this works for matlab.
$ cp maple.linux.v10.installer.bin maple.linux.v10.installer.bin.bak
$ cat maple.linux.v10.installer.bin.bak | sed "s/export LD_ASSUME_KERNEL/#xport LD_ASSUME_KERNEL/" > maple.linux.v10.installer.bin
$ ./maple.linux.v10.installer.bin
Offline
Thanks for the suggestion, but:
[davide@hp-arch cd1]$ cat install | grep -c LD_ASSUME_KERNEL
0
and it's the same with every file I have checked... In particular I have checked the ones in the $CD2/update/pd/toolbox/symbolic that seems to be the ones linked with the symbolic toolbox and maple...
Offline