On Thu, Oct 24, 2019 at 10:41:49AM -0700, Bart Van Assche wrote:
> On 10/24/19 10:27 AM, Omar Sandoval wrote:
> > On Mon, Oct 21, 2019 at 03:57:18PM -0700, Bart Van Assche wrote:
> > > +# System uptime in seconds.
> > > +_uptime_s() {
> > > + local a b
> > > +
> > > + echo "$(</proc/uptime)" | {
> >
> > What's wrong with cat /proc/uptime? Or even better,
> >
> > { read ... } < /proc/uptime
>
> Hi Omar,
>
> As you probably know 'cat' triggers a fork() system call but echo $(<...)
> not. This is a performance optimization. Input redirection would also work.
>
> > > + read -r a b && echo "$b" >/dev/null && echo "${a%%.*}";
> >
> > What's the point of the echo "$b" here?
>
> That echo "$b" statement suppresses a shellcheck warning about $b not being
> used.
>
> > Seems like this could all be condensed to:
> >
> > { read -r s && echo "${s%%.*}" } < /proc/uptime
> >
> > But that's more cryptic than it needs to be. Can we just do:
> >
> > awk '{ print int($1) }' /proc/uptime
>
> That's a valid alternative, but an alternative that triggers a fork() system
> call. I don't have a strong opinion about which alternative to choose. Do
> you perhaps have a preference?
The awk option is more readable, and we're not trying to win any
performance awards in blktests. Let's just go with that.
