Code Duplicated in Multiple Languages

[djkees]

Hello!

I've been looking into the unit conversion functions and universal constants, and noticed they're implemented separately in both Python and Fortran (at least those two; I haven't checked the other bindings yet). Atomic weights are the exception -- Python calls straight through to the Fortran atomic_data.f90 rather than maintaining its own copy.

That inconsistency in approach seems to have led to real drift: for example, source/units.f90 hardcodes the mmHg-to-bar conversion as 0.001333 (rounded), while the Python side (source/bind/python/cea/units/init.py) computes a more precise 0.0013322368... -- about a 0.017% relative difference. Small today, but it's the kind of thing that compounds as more constants/conversions get duplicated across languages.

Before I put time into fixing individual files, I wanted to ask about the intended structure: should Python bindings reimplement this kind of logic independently, or call through to Fortran as the single source of truth, the way atomic weights already do?

I checked the documentation site and didn't see guidance either way, but I may have missed it. Thanks!

@markleader

[markleader]

I am in favor of keeping the unit conversions separate across Fortran/Python, but I am supportive of updating them to the same values, where more precision is preferred.

[djkees]

Thanks! Does that hold for things like atomic weights too? Would you prefer if an atomic weights file was made for Python?