reference_point can't find ifgramStack.h5

[lghackett]

I am running the smallBaselineApp on a stack of Sentinel-1 interferograms produced by ISCE, and am running into a strange bug where the reference_point step cannot find the inputs/ifgramStack.h5 file, despite the fact that I can confirm it (i) exists (ii) is uncorrupted, and (iii) is accessed in previous steps, as you can see in the attached images.

The error:

First, see that the previous steps generating the file run without hiccups, and that other steps are able to access the file fine.

Now, see that accessing this file directly using ipython also works just fine:

Trying to run reference_point separately does not help:

Any clues on what I can do to make the algorithm keep going?

Other details:

• Working on Mac OS 14.7.5
• MintPy version 1.5.3

[ditafaith]

@lghackett I suggest you to isolate the error by make it more relax, for example, no need to set up the reference point first. Let's see it.

[lghackett]

Thank you for your response! If I try to go to the next step I get the same error again:

[ditafaith]

Please make sure to clean all files (exclude the config file) before run the smallbaseline.
Also, what is your reason keep 1.5.3 rversion rather than the latest one 1.6.1?

[lghackett]

I had reverted to 1.5.3 because I was having other HDF5 issues with the newer version.

I don't understand why exactly this worked, but the problem was fixed by switching to a different computer. Given that the conda environment was the same between the two computers (now using version 1.6.1), I think the problem must have something to do with my computer's ability to deal with the HDF5 files / compression (OS? Developer tools version?)

Here are the specs of the computer that was able to run the program in case anyone else runs into this:

Sucessful on: MacOS 13.7.5
conda environment created with:

mamba create -n mintenv
mamba install isce2 mintpy

Thank you again for your responses @ditafaith !!!