3 ms·
What exactly is the call to action here? For kernel developers to change how time works, and probably break thousands of drivers and applications, based on a bl
by brianpgordon 13y ago
What exactly is the call to action here? For kernel developers to change how time works, and probably break thousands of drivers and applications, based on a blog post?
- Udo 13y agoNo, willful misunderstanding aside, the idea is to provide a single continuous counter to apps that want it. If you look closely at most implementations, that's how many apps actually expect it to work.
- jonhohle 13y agoThere already is a continuous monotonic timer available in most operating systems, it just happens to be relative to some internal counter local to that system (and depending on virtualization layer doesn't always provide monotonic guarantees). This value can generally be translated into a fixed (and vice-versa), serialized timestamp for persistence.
- AndrewBissell 13y agoI think that's mainly a problem with apps using a wall clock time (e.g. System.currentTimeMillis() in Java) when what they really want is a monotonic clock based on a hardware counter of some kind (like System.nanoTime()). I'm not too familiar with what the clock situation is like for languages outside C/C++/Java.