Ukko, 4 months ago

@nyoom this triggers the UB fear inside of me, this feels like it'd act differently on a big enhian system

nyoom, 4 months ago

@Ukko yeah, I'm not quite sure why reinterpret isnt more clearly marked as unsafe the way most things that rely on memory layout in julia are. maybe because it can't cause memory corruption. I was just using reinterpret and sizeof to inspect the bit layout

(speaking of, that gives me an idea for a simple helper tool...)

Ukko, 4 months ago

@nyoom what would happen if you try to reinterpret, eg, array of 3 int64s into an array ef structs each 2 int64s?

nyoom, 4 months ago

@Ukko it throws an error

Ukko, 4 months ago

@nyoom ok that's good i guess

nyoom, 4 months ago

@Ukko I mean it has to really given that arrays are bounds-checked by default, so it needs to store the size of the resulting array somewhere

Ukko, 4 months ago

@nyoom an alternative could've been padding everything with zeroes or dropping all the data not fitting in the struct

nyoom, 4 months ago

@Ukko yeah, though that'd be confusing for multidimensional arrays ig

nyoom, 4 months ago

@Ukko done, this should dump any isbits type into a hex code