tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: preserving full timestamps



On Thu, Nov 27, 2014 at 04:28:26AM +0000, David Holland wrote:
> On Thu, Nov 27, 2014 at 09:47:22AM +0900, tsugutomo.enami%jp.sony.com@localhost wrote:
>  > As you know, userland tools like cp(1) currently preserve timestamps
>  > only up to microseconds order since there was no system call to set full
>  > timestamps.
>  > 
>  > Nowadays, we have such system calls.  So, I'd like to change those tools
>  > to set full timestamps unless there is an objection.
>  > 
>  > In addition, test(1) is already changed to compare full timestamps.  So,
>  > unlike on netbsd-6 or earlier, currently the result of test -ot/nt may
>  > be inconsistent even when timestamps are copied in a same filesystem or
>  > between filesystems of same type.
> 
> Sounds like a good idea to me...
> 
> (FWIW, I don't think any of it should be #ifdef __NetBSD__; that just
> causes it to misbehave silently when compiled on another platform. But
> it's not a big deal.)

That depends primarily on whether it is also code used in src/tools.
Code that doesn't get build as tool AND doesn't have a portable version
should not use __NetBSD__.

Joerg


Home | Main Index | Thread Index | Old Index