Re: We need a new rendering engine written from the ground up
The trouble is that foresight is not 20/20. Things change. For example, someone who wired up a place with for 10/100 Ethernet suddenly finds a problem trying to set up for 10Gig Ethernet because the necessary cabling for it didn't exist at the time. There was no way to future-resist something like that because the future technology was past the black wall of uncertainty. Eventually, the cost to upgrade piecemeal (due to lack of scale and continual bodging) approaches the cost to replace, at which point (like a car that would cost more to repair than to replace) it's better to start fresh.
Incremental changes can lead to bloat if they keep building up. Put it this way. Recall that Firefox at its beginning was a second system: a rewrite of Netscape. Then it became a lot of incremental changes built up over time. Sort of like the unsorted heap of clothes. Eventually, the time comes to "clean up shop". Someone hinted at a lot of bodge-work and unused code. I recall an article about a hole in QEMU that involved the very-old-but-still-active floppy emulator. Sure, don't fix what isn't broken, but sometimes things can be broken without your knowledge.
Biting the hand that feeds IT
