[kepler-users] RExpression Bug?

Tom Parris parris at isciences.com
Thu Dec 3 14:01:06 PST 2009


Dear Keplerites,

I believe I have stumbled across a bug in the RExpression actor.  I am using
the 1.0.x developer version from svn in Windows XP (x86).

The attached workflow should demonstrate two bugs.

1. Matrices get scrambled by RExpression.  The "Make a Matrix" actor is a
simple python script that creates an nrow x ncol matrix where M[i,j] = i+j
(with subscripts starting at 0).  The result is displayed properly by
MatrixViewer and MatrixViewer2.  However, the matrix is scrambled inside R
(see Display and ImageJ).  It looks like the row and column subscripting is
reversed.

2. The workflow executes the first time the run button is pressed, but
produces a type error the second time it is pressed.  Reloading the workflow
solves the problem (for one run iteration).

With regards,
Tom

p.s.  There is also a serious performance issue.  Try changing the nrow and
ncol constants from their current values of 5 and 10 to something like 100
and 200.  Even with the extra conversions from binary to text and back, it
takes way too long.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nceas.ucsb.edu/kepler/pipermail/kepler-users/attachments/20091203/4f1cca43/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: RExpression_Matrix_IO_Test_V2.xml
Type: text/xml
Size: 46092 bytes
Desc: not available
URL: <http://lists.nceas.ucsb.edu/kepler/pipermail/kepler-users/attachments/20091203/4f1cca43/attachment.xml>


More information about the Kepler-users mailing list