Do not use relpath?

[floriankrb]

This use of relpath :

cfgrib/cfgrib/messages.py

Line 552 in 3a53b3a

return os.path.relpath(self.fieldset.path)

is causing an issue on window when path and start are on different drives.

Indeed in the os.path.relpath documentation:

Return a relative filepath to path either from the current directory or from an optional start directory. 
This is a path computation: the filesystem is not accessed to confirm the existence or nature of path or start. 
On Windows, [ValueError](https://docs.python.org/3/library/exceptions.html#ValueError) is raised 
when path and start are on different drives.

This happens in real use cases:

https://github.com/Climdyn/climetlab-eumetnet-postprocessing-benchmark/runs/5217438193?check_suite_focus=true#step:8:508

ValueError: path is on mount 'C:', start on mount 'D:'

Perhaps a try/catch would help here?

[iainrussell]

Hi @floriankrb,

So what's the correct thing to do here? Something like this?

def source(self) -> str:
    try:
        return os.path.relpath(self.fieldset.path)
    except ValueError:
        return self.fieldset.path

[alexamici]

@iainrussell your fix is one possibility, the other option is to return "N/A".

The idea after the relpath is that the source is written in the history attribute and it may end up leaking the folder structure of the machine (possibly including the user home folder and username) in a netCDF file. This is especially a problem in production environments.

[floriankrb]

Yes, I think that any of these two solutions would work.

I would go for what @iainrussell proposed, based on the (very biased) assumption that if you use windows in a production environment, security and information leak are not your main concern.

[iainrussell]

Actually, after reading what @alexamici wrote, I think I'm more inclined to return basename(path) in this case, so we get the filename of the GRIB file. In many cases this is probably what relpath(path) would return anyway.