@bmattlife@rep_movsd@BHolmesDev instead of jumping to "crash the entire program on overflow" there could be language constructs that explicitly handle overflow at the block level, e.g. some kind of calculate{ ... }on_overflow{ ... }.
@graninas I don't want to have N copies of libraries for the N programs that use those libraries. I want one copy of each library and the N programs that use it should use that one copy.
@davidtgoldblatt@redneckleeusmc Are there any situations where these optimizations are actually useful at all? What are projects that compile with -fno-strict-alias leaving on the table?
@myrrlyn it would be more efficient to allocate a single block of FRAMES*CHANNELS*H*W*sizeof(int) then cast it to an int(*)[CHANNELS][H][W] The code in screenshot is correct but not efficient.