On 2010-12-26 15:40-0800 Alan W. Irwin wrote: > The current status is I am expanding the list of octave examples to be > tested to include the few more beyond standard example 10 which do not > use array arguments.
It turns out every example other than 10 has some call to a PLplot function that has an argument that uses one of the current placeholder typemaps for octave. So expansion to more examples will have to wait until those placeholder typemaps get defined properly. I have also streamlined and corrected the build-system logic for the -DENABLE_swig_octave=ON case. The result is to use swig-generated octave bindings rather than matwrap-generated ones, all you need to do is specify -DENABLE_swig_octave=ON, and the build system does the rest. Here are the current results on the Linux platform for the test_diff_psc target in that case: octave Missing examples : 01 02 03 04 05 06 07 08 09 11 12 13 14 14a 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 33 Differing postscript output : Missing stdout : Differing stdout : which shows that all but 10 is missing (as expected since only example 10 is currently run for this case), but the results for that example are perfect (no diffs). There are three issues I am aware of for the swig-generated octave bindings as a result of my Linux platform tests. 1. An extremely minor issue is that the plplot_octave command needs to be wrapped as "plplot_stub" so those using the swig bindings (once those are completed) will still be able to use the plplot_stub command to gain access to plplot from octave. (I currently avoid this issue via configuration of test_octave.sh. That takes care of the issue for the automated testing environment, but such configuration does not help interactive users.) 2. A less minor issue, is you cannot use the gcc option -fvisibility=hidden on Linux (I don't use this option in any case for MinGW/MSYS/wine) because that hides plplot_octave completely (and silently except for a puzzling plplot_octave undefined error which took me quite a while to track down). I believe this must be due to a bug in swig octave support because we have no visibility issues like this in our swig-generated Python, Java, and Lua bindings. 3. The important issue is that the placeholder typemaps need to be worked on in bindings/octave/plplot_octave.i so that the entire PLplot API becomes accessible to the swig-generated octave bindings which will allow running all the examples. These typemaps should be straightforward to fix since the code generated by matwrap is available as a template of what to do for each kind of PLplot API argument. I have ignored these issues for now and have moved on to the wine build and test of our swig-generated octave bindings. That is looking quite promising. The build got quite far before an issue was encountered concerning finding the hdf5.h header file. I will deal with the necessary build-system tweak after I get some sleep. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------------ Learn how Oracle Real Application Clusters (RAC) One Node allows customers to consolidate database storage, standardize their database environment, and, should the need arise, upgrade to a full multi-node Oracle RAC database without downtime or disruption http://p.sf.net/sfu/oracle-sfdevnl _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
