I really wonder why people are so eager to "completely get rid of the
old /dev/random and /dev/urandom code" when
1. the only defects that have been identified so far require only
minor tweaks to that code,
2. the code has been around for some time and there's some reason to
believe it's pretty solid,
3. apart from the issues under (1), the proposed replacement (yarrow)
uses essentially the same design principles,
4. the proposed replacement either requires vast amounts of work to
fit into Linux (yarrow) or is something else that hasn't been defined
at all yet.
In other words, why propose doing large hunks of work to fix small
issues?
Of course, if you want to do X and don't mind that you defined X to be
large, that's your choice. But if you're trying to encourage someone
else to do X, it's best to make X as small as is possible while still
getting the job done.
Am I missing something here? If so, what is it?
paul
