Rplot001.png left around when doctesting

[was]

I doctested after I think #7665 and this file Rplot001.png appeared in SAGE_ROOT. That's obnoxious.

[kcrisman]

eah, this is my fault:

+        This example saves a plot to the standard R output, usually 
+        a filename like ``Rplot001.png`` - from the command line, in 
+        the current directory, and in the cell directory in the notebook::
+
+            sage: r.plot("1:10")
+            null device 
+                      1 
+

I actually put this one in to show users how to actually get a file in a directory that they have access to, as opposed to a temp directory, but forgot we can't do that. Is it legitimate to put #not tested with this doctest as a fix, since it is important to have the example there?

[kcrisman]

but forgot we can't do that. Is it legitimate to put #not tested with this doctest as a fix, since it is important to have the example there?

From was:

Another option:

1. make sufpre the png file you are about to write doesn't exist in the current directory, then write it.
2. then delete it!


# not tested should be avoided at all costs.

[kcrisman]

Based on Sage 4.4 + R plot patch

[kcrisman]

Okay, try this - it is VERY explicit. I also replaced occurrences of os.unlink with os.remove, which is identical but makes far more sense to those who are not familiar with the unlink command (like me).

[was]

I really don't like this very brittle and dangerous patch. I've posted a new patch. Please review. I've included this in alpha3 anyways.

William

[mvngu]

The directory named by the environment variable SAGE_TMP is where junk files can go. Such junk files include temporary files generated during doctesting.