EDIT: There is a guide for interoperating GILDAS and MIRIAD here : http://www.iram.fr/~gildas/gildas-miriad/ 

I am now trying to work with data from the Submillimeter Array (SMA), calibrated in IDL with the MIR package, continuum subtracted in MIRIAD, exported to UVFITS and read in to GILDAS for imaging. Yes, it is a mess.

It kind of works. Except the RMS noise somehow is 10-15% higher for the imaged data in GILDAS than for the same uv-data and same RMS channels in MIRIAD. Why? I do not know, but I have now resorted to work only in MIRIAD with the MIR calibrated data. I will conduct my normal analysis there, and output a FITS file of the cleaned data to work with my own scripts for creating various maps and spectra.

The best would be if I could write/use some Python modules for fitting Gaussians in the uv-plane as well, so I could finally break free of the dependence of MIRIAD/GILDAS/CASA/whatever package that is used at telescope X.