Is there any downside of using file names with \setmainfont, except for the fact that it takes more typing?
I ask because I am experiencing problems with using names for TeX Gyre fonts on my Linux Mint box: as explained here, XeLaTeX on Linux will not find TeX-installed fonts by name. Hence, I installed the fonts also system-wide (Ubuntu package tex-gyre), but the font names are different from my Windows box: the spaces are gone, so "Tex Gyre Termes" becomes "TeXGyreTermes" etc. This means that using names, the code becomes system-dependent :-( (By the way, should this be considered a bug and reported to the maintainers of the tex-gyre package?)
On the other hand, calling the font by file name should work as long as the font file names are the same - can one trusts that? And is there any other potential problem with using file names?