HDF and H5 files in the same directory

[mbennun]

If there are H5 and HDf files in the same directory (both from the same run/time steps) PsiPy will read both of them which is problematic of course.

[dstansby]

Oops that doesn't sound good! I'll see what I can do. Out of interest what happens? A crash?

[mbennun]

No it does not crash. It reads everything but there seems to be junk, NaN everywhere

[mbennun]

A related issue is the following: Sometimes we create cuts of the HDF files or manipulate them and write the results in the same directory, for example I have this in one directory:
b_ratio.h5 bp_rp.h5 br_1d.h5 bt_1d.h5 hmasip01 rho_rp.h5 vr002.h5 vt_rp.h5
bp_1d.h5 br002.h5 br_rp.h5 bt_rp.h5 rho002.h5 vp_rp.h5 vr_rp.h5

PsiPy reads all the files and this messes things up..

[dstansby]

When there are both .h5 and .hdf files present for the same variables in the same folder, what do you think should happen? Raise an error? Or read one set of files, but warn about the duplicate files?

For the cuts, is there a clear naming structure for the cut files that is different from the naming structure for the full 3D (original) files?

[thepeteriley]

When there are both .h5 and .hdf files present for the same variables in the same folder, what do you think should happen? Raise an error? Or read one set of files, but warn about the duplicate files?

an optional parameter argument specifying hdf_type = 'h5' or 'hdf' would allow the user to specify. Then it would be their responsibility.

For the cuts, is there a clear naming structure for the cut files that is different from the naming structure for the full 3D (original) files?

Unfortunately not. If they are BC files, then yes "br_r0.hdf", etc., but if a user has created 2D files, they can be anything. Limiting the readable files to 001 to 000001 formats should address this, and if we add lower-level routines to read in the HDF files that would provide a way for all files to be read.