The Closure Compiler does a fine job of replacing all such expressions with their decimal equivalent, relieving us from having to calculate them ourselves, and eliminating any runtime impact
There are still a few regressions to address, but keyCode-to-simCode-to-scanCode conversion feels more robust/cleaner, and new features like programmable repeat rates will actually be doable now
Some other interesting bugs were fixed along the way:
the Video component incorrectly assumed that masking a negative number with 0x80000000 would yield 0x80000000 (it actually yields -0x80000000, thank you JavaScript);
the ChipSet component provides a new addCMOSMemory() interface that RAM objects can use to bump the CMOS total to match the actual total;
the CPU properly implements INSW and OUTSW instructions now (kind of important for the AT hard disk controller)
