2012/10/4 José Matos <jamatos(a)fc.up.pt>:
On 10/04/2012 07:23 AM, Vedran Miletić wrote:
> I haven't tested it thoroughly, but it works. Running sage installed
> via a package manager gives a good feeling indeed.
>
> Thanks for your effort.
>
> Regards,
> Vedran Miletić
I agree, thanks for your persistence Paulo. :-)
:-)
I just updated the f17 repository to provide sagemath 5.4.beta1
There is also an updated numpy package with the patches from
https://github.com/numpy/numpy/issues/291
For rawhide there is not yet a proper repository, but besides
also providing a pre built numpy (that should not be required
soon, as the patches are in the main numpy branch), I also
added fplll patches, that I also posted at
http://trac.sagemath.org/sage_trac/ticket/12835
One issue that was working flawlessly previously was
the jmol interface, that now is not fully functional, I got it
to work by using the JmolData.jar in the jmol tarball, but
that is cheating as it would use a prebuilt jar. I almost got
it to work by patching like this:
---%<---
diff -up sage-5.4.beta1/spkg/build/sage-5.4.beta1/sage/interfaces/jmoldata.py.orig
sage-5.4.beta1/spkg/build/sage-5.4.beta1/sage/interfaces/jmoldata.py
--- sage-5.4.beta1/spkg/build/sage-5.4.beta1/sage/interfaces/jmoldata.py.orig 2012-10-20
13:09:00.932379141 -0300
+++ sage-5.4.beta1/spkg/build/sage-5.4.beta1/sage/interfaces/jmoldata.py 2012-10-20
16:39:46.791863395 -0300
@@ -86,7 +86,7 @@ class JmolData(SageObject):
sage_makedirs(jmolscratch)
scratchout = os.path.join(jmolscratch,"jmolout.txt")
jout=open(scratchout,'w')
- testjavapath = os.path.join(SAGE_LOCAL, "share", "jmol",
"testjava.sh")
+ testjavapath = os.path.join(SAGE_LOCAL, "bin",
"testjava.sh")
result = subprocess.call([testjavapath],stdout=jout)
jout.close()
if (result == 0):
@@ -185,7 +185,7 @@ class JmolData(SageObject):
scratchout = os.path.join(jmolscratch,"jmolout.txt")
jout=open(scratchout,'w')
#now call the java application and write the file.
- result =
subprocess.call(["java","-Xmx512m","-Djava.awt.headless=true","-jar",jmolpath,"-iox","-g",sizeStr,"-J",launchscript,"-j",imagescript],stdout=jout)
+ result =
subprocess.call(["java","-Xmx512m","-Djava.awt.headless=true","-classpath","/usr/share/java/Jmol.jar:/usr/share/java/vecmath.jar:/usr/share/java/apache-commons-cli.jar","org.openscience.jmol.app.JmolData","-iox","-g",sizeStr,"-J",launchscript,"-j",imagescript],stdout=jout)
jout.close()
else:
errStr = "Java Virtual Machine not available.\n"
---%<---
but it was failing with a null pointer exception when loading
the interface to export a png image, and I could not get
enough backtrace information to try to figure out what jar
file was missing, and preventing it from generating a
"quick" thumbnail of the plot...
It would also require rebuilding jmol anyway, to either build
JmolData.java and add JmolData.class to Jmol.jar, or install
JmolUnsignedF.jar as JmolData.jar.
So, for now the jmol interface wil show a harmless error
message (due to not installing testjava.sh actually), and
have a button to start an interactive plot. The new "advanced
controls" work.
--
José Matos
Paulo