Question about TIMEOUT applet.

[Bernhard Reutner-Fischer]

On 15 July 2013 23:25:53 Harald Becker <ralda at gmx.de> wrote:
> Hi Denys !
>
> >> > (at least newer) coreutils package contain a TIMEOUT command
> >> > with a slightly different syntax then Busybox. Is this
> >> > intentional or a bug? What about upstream compatibility?
> >> No, it is not intended. This needs to be fixed.
> >
> >It seems to me the safest fix would be to keep the -t option,
> >and only interpret the first non-option argument as the timeout
> >value if the -t option was not specified on the command line.
> >This would make the busybox version accept both its old syntax
> >and the coreutils syntax.
>
> I stated looking into this and found it would be best to create a
> libbb function to get a DURATION argument with possible suffixes.
> This can also be used for the sleep applet.
>
> The libbb function shall depend on CONFIG_FEATURE_FANCY_SLEEP to
> enable usage of suffixes (else only seconds), and on
> CONFIG_FEATURE_FLOAT_SLEEP to allow fractionals (else only
> integers). The functions returns the given duration in a struct
> timespec.
>
> Beside this I like to move the timeout applet from
> miscutils to coreutils as upstream timeout is provided in this
> package.
>
> Is this ok for you?

I would unconditionally support floats in the factored-out helper. The rest 
sounds OK.
Thanks,


Sent with AquaMail for Android
http://www.aqua-mail.com