@fsouchu I see ^ _ ^;
It is very unlikely that fetch will support partial reads in the future -- I view the files themselves as the atomic chunks that are part of a larger thing (cartridge). Would it be hard to break that data up into pieces? I'm planning on enforcing RAM limits for 0.1.1 but will add a way to measure it before then.
It might also turn out there is a way to abuse /ram, but that also needs to have some kind of limit..