Re: The fate of libc5
>>>>> "Ariel" == Ariel Tankus <arielt@math.tau.ac.il> writes:
Ariel> Unfortunately, MATLAB is compiled against libc5, and the
Ariel> official requirements in MathWorks(TM) site is :
Ariel> Linux 2.0.34 kernel - Red Hat 4.2, 5.1 Debian 2.0 (as
Ariel> appears in:
Ariel> http://www.mathworks.com/products/matrix/linux.shtml).
Ariel> The reason it matters how MATLAB was compiled for the
Ariel> end-user is that when one write MEX-files, one has to use a
Ariel> compiler which... uses libc5.
I was hoping somebody would bring this up (I wasn't sure of my
sources). Do up-to-date versions of Matlab support libc6? I guess
not.
Matlab is also a reason to keep libc5 development files (not that I
use them myself). I think you can convert Matlab scripts into C code
(is this anything to do with MEX?), which rely of good old libc5
libraries :-(.
However, I do not understand the difficulties in maintaining
development support for libc5, I get the impression it might be
difficult.
Ariel> As someone who mainly uses MATLAB in work, it is a real
Ariel> reason to keep libc5 compatability libraries, despite the
Ariel> headache it probably causes.
I know somebody who can't use Redhat for this reason (I get the
impression Redhat doesn't support libc5 anymore), but has to use SunOs
instead.
Personally, I would try to use octave instead (claims to be a mostly
Matlab compatible), which is in main (it is DFSG free), but suspect it
doesn't have some of the GUI signal processing features I have seen in
Matlab (might be an add-on). I dislike any software that still uses
libc5 simply because it looks like the authors aren't serious about
maintaining it.
As I said, I am not sure of my sources, and may have some of the
above wrong (I assume Redhat 4.2 and 5.1 is libc5 based?).
--
Brian May <bam@debian.org>
Reply to: