> It would be far nicer to have a single implementation that was known to work so that kernel developers could more profitably use their time introducing bugs i

> It would be far nicer to have a single implementation that was known to work so that kernel developers could more profitably use their time introducing bugs into code that is truly unique to their problem area. Emphasis mine. That was a fun turn of phrase.