Installing SAGE 4.3.1 on Fedora 12 with x86_64

[A. Jorge Garcia]

I've used http://www.sagenb.org extensively in my AP Calculus BC
classes. I use the liveCD at just about every conference I attend.
In fact, I'm giving a presentation tomorrow with the liveCD to entice
students to sign up for a new Calculus Research Lab for next year.

So, sorry for being dense, but I must be missing something fundamental
here when trying to install SAGE to a Linux partition for my Calculus
students. I have a LAN of HP PCs with AMD dual core 64 bit Athlons
where I have installed the Fedora 12 DVD for my AP Computer Science
classes. Silly me, I decided to download the SAGE 4.3.1 Fedora 12
x86_64 binary and install it. Well the download went fine, the md5
checked out and it seems to have extracted correctly to my /home
dir.

So, I tried running SAGE in a terminal from the SAGE dir that was
created. I got a welcome message about typing notebook() to start the
GUI and then some sort of error message about not being able to run an
executable. It looks like it was in the bin/python dir when it
crashed. Anyone know what went wrong here and how to fix it?

One thing comes to mind. The download page (Harvard?) said something
about apt-getting gfortran before running SAGE. Well, that's a Debian
command, so I tried "yum install gfortran" but Fedora didn't find it.
I was hoping that gfortran was only needed when compiling from source,
not when running the binaries....

Please help,
A. Jorge Garcia
http://calcpage.tripod.com

[Dan Drake]

On Thu, 04 Feb 2010 at 05:16PM -0800, A. Jorge Garcia wrote:
> So, I tried running SAGE in a terminal from the SAGE dir that was
> created. I got a welcome message about typing notebook() to start the
> GUI and then some sort of error message about not being able to run an
> executable. It looks like it was in the bin/python dir when it
> crashed. Anyone know what went wrong here and how to fix it?

If you post a log of what happened, we might be able to figure out what
went wrong. :)

> One thing comes to mind. The download page (Harvard?) said something
> about apt-getting gfortran before running SAGE. Well, that's a Debian
> command, so I tried "yum install gfortran" but Fedora didn't find it.
> I was hoping that gfortran was only needed when compiling from source,
> not when running the binaries....

Actually, I believe that now you do need gfortran just to run the
binaries. It's not the compiler that you need, but rather the shared
library that comes along with it.

I googled up "redhat gfortran" and I think "yum install gcc-gfortran"
will install what you need.

Dan

--
--- Dan Drake
----- http://mathsci.kaist.ac.kr/~drake
-------

[William Stein]

On Thu, Feb 4, 2010 at 7:41 PM, Dan Drake <dr...@kaist.edu> wrote:
> On Thu, 04 Feb 2010 at 05:16PM -0800, A. Jorge Garcia wrote:
>> So, I tried running SAGE in a terminal from the SAGE dir that was
>> created.  I got a welcome message about typing notebook() to start the
>> GUI and then some sort of error message about not being able to run an
>> executable.  It looks like it was in the bin/python dir when it
>> crashed.  Anyone know what went wrong here and how to fix it?
>
> If you post a log of what happened, we might be able to figure out what
> went wrong. :)
>
>> One thing comes to mind.  The download page (Harvard?) said something
>> about apt-getting gfortran before running SAGE.  Well, that's a Debian
>> command, so I tried "yum install gfortran" but Fedora didn't find it.
>> I was hoping that gfortran was only needed when compiling from source,
>> not when running the binaries....
>
> Actually, I believe that now you do need gfortran just to run the
> binaries. It's not the compiler that you need, but rather the shared
> library that comes along with it.

This will change for the next Sage release, by the way. If nothing
else, I'm going to just copy the gfortran lib into local/lib/ when
creating the binaries on the build machines. However, -bdist or
sage_fortran should I think do this automatically.

-- William