#2245 Merged at 1d70bf4
  1. MattT

This changes the main pixelization routines to accept input buffers.

One of the biggest impediments to multi-block systems is the way that pixelizers currently allocate their own buffers. This allows them to receive input buffers.

I have changed, likely incorrectly, rows/cols in some of these pixelizers (cc @Yi-Hao Chen ) to be more consistent.

While I think the "work" here is done, this isn't ready to go in until after feature freeze is over and until after Yi-Hao and I work through the rows/cols issues.

UPDATE 1: Merging with upstream, and correcting my confusion about who issued the row/column PR. Many apologies.

  • All tasks resolved

Comments (35)

  1. Nathan Goldbaum

    Using yi-hao's test script from issue 2120, it looks like there's still some inconsistency between the number of pixels along the x and y directions:

    import yt
    ds = yt.load("GasSloshing/sloshing_nomag2_hdf5_plt_cnt_0150")
    slice = yt.SlicePlot(ds, 'x', "density", width = (100.0, 'kpc'))

    Which produces: http://i.imgur.com/mp770iB.png

    I would expect the image to have 8 pixels along the x direction and 4 along y.

  2. Yi-Hao Chen

    Thanks for working this out! It would be useful for people having non-square domains. At first glance, it all looks good. I will take a deeper look when I have time.

    • Add tests for cylindrical pixelization
    • Deprecate CylindricalFixedResolutionBuffer
    • Add tests for geographic pixelization
    • Add non-unitary aspect ratio tests for non-cartesian pixelization
    • Add tests for spherical pixelization
  3. MattT author

    I've merged and responded to a couple comments, but I'm going to be issuing a separate PR to add tests for the pixelization calls (not plot window) for cylindrical and spherical meshes. Then those answers can be generated, and these compared. @Yi-Hao Chen has one point that I'm not sure about (I have tested and I get the same results as before the PR for the test cases I've checked) but otherwise, I think once the tests go in this will be ready to go in.

    If I can get this one reviewed, then once the forthcoming one goes in, I'll have this one ready.

  4. Nathan Goldbaum

    It looks like merging in my annotate_cell_edges PR triggered some conflicts. I'm happy to help out with that if you need a hand - it was sensitive to row/column ordering so it might be non-trivial.

  5. yt-fido