My responses are imbedded. On Oct 22, 2014, at 5:40 PM, "Guyer, Jonathan E. Dr." <[email protected]> wrote:
> > On Oct 22, 2014, at 2:48 PM, Seufzer, William J. (LARC-D307) > <[email protected]> wrote: > >> After attempting to save a VTK formatted file while running on a cluster... >> I now see in the FAQ that the VTKCellViewer must have a display. > > I'm not saying this is wrong, but where did you find this? I don't find any > such discussion in our FAQ (and I'm the most likely one to have written it). > While it is not stated directly, it is easy to imply from this... What if I only want the saved file, with no display on screen? To our knowledge, this is only supported by Matplotlib, as is explained in the Matplotlib FAQ on image backends. Basically, you need to tell Matplotlib to use an “image backend,” such as “Agg” or “Cairo.” Backends are discussed athttp://matplotlib.sourceforge.net/backends.html. When I attempt to create a VTK viewer, while running on the cluster, I get a "Cannot connect to X server" error from each core while on the cluster. I'm looking into the possibility of installing Xvfb (a virtual frame buffer) as a workaround. >> This is not possible running in parallel on a queued cluster. > > Assuming this restriction is accurate, it may stem from our use of tvtk. This > made sense at the time, but is an issue when trying to build a stack on > anaconda, for instance. It would be better, I think, if we used PyVTK. I don't know if the issue would go away with PyVTK since it might also need a frame buffer. > >> I have also attempted to dump (pickle) cell variables, but a post processing >> single core script is unable to read the dump file (tried a few months ago, >> forgot the error). > > Looking back at your post from August, I think Daniel fixed this issue > > http://thread.gmane.org/gmane.comp.python.fipy/3550/focus=3562 > > Currently available (and waiting for me to merge) here: > https://github.com/wd15/fipy/tree/issue412-pickle_comm > > Can you try again? Yes, I'll give this another try. For my purposes, mainly visualization at this point, saving to a VTK file appears to be a good approach. I can directly load the file into Mayavi without any need for FiPy code to un-pickle the CellVariable and then pass it to a Mayavi viewer. I've been successful, on my laptop, writing a cell variable to a vtk file and displaying it in Mayavi. > > >> Could you suggest a work-around for getting 3D cell data into Mayavi -AFTER- >> my fipy code has run? > > As I commented when you posted a few months ago, I *really* need to finish up > my Xdmf branch. It's quite out of sync with current development, but it > should be largely self contained and visible at > https://github.com/usnistgov/fipy/tree/xdmf/fipy/io/xdmf. It might only work > for Grid3D. > _______________________________________________ > fipy mailing list > [email protected] > http://www.ctcms.nist.gov/fipy > [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ] Cheers, Bill
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ fipy mailing list [email protected] http://www.ctcms.nist.gov/fipy [ NIST internal ONLY: https://email.nist.gov/mailman/listinfo/fipy ]
