Subject: Re: [ast-users] Discipline functions for float vs. $ printf "%f"
s_t.varname # ...
--------
> Hi!
>
> ----
>
> The following testcase contains a simple "stopwatch" type which tracks
> the time between the functions "start" and "stop" and maintains a
> variable "diff" which contains the difference:
> -- snip --
> typeset -T stopwatch_t=(
> float starttime=nan
> float stoptime=nan
> float diff=nan
>
> function start
> {
> (( _.starttime=SECONDS ))
> return 0
> }
> function stop
> {
> (( _.stoptime=SECONDS ))
> return 0
> }
> function reset
> {
> (( _.starttime=nan , _.stoptime=nan ))
> }
> function diff.get
> {
> (( .sh.value=_.stoptime-_.starttime ))
> print -u2 "#mark"
> return 0
> }
> )
>
> stopwatch_t sw
>
> sw.start
> sleep 2.1
> sw.stop
>
> printf "time=%f\n" sw.diff
> -- snip --
>
> However when I run this testcase with ast-ksh.2009-06-22 it only prints:
> -- snip --
> time=nan
> -- snip --
> ... while the expected output should AFAIK be:
> -- snip --
> #mark
> time=2.1000
> -- snip --
>
> AFAIK two things go wrong there:
> 1. Using $ printf "time=%f\n" sw.diff # to call the discipline method
> stopwatch_t.diff.get does not work, however it works when I replace
> "sw.diff" with "${sw.diff}"
The get discipline is only called to get string values, $var. Internally,
the shell uses a different discipline to get a numerical value.
I will consider adding a getn and setn disciplines which will allow an
assignment to .sh.value which will be type float to a future release.
> 2. The somehow the value assigned to ".sh.value" does not make it to the
> "printf" builtin... but I am not sure why this happens...
The other problem is that in diff.get, The _ variable is set to sw.diff,
not sw. Therefore, you can change function diff.get
to
function diff.get
{
nameref __=${.sh.name%.*}
(( .sh.value=__.stoptime-__.starttime ))
}
Better you, when not just define
diff()
{
(( _.stoptime - _.starttime ))
}
which would allow
printf "time=%f\n" ${sw.diff}
to print the time.
>
> ----
>
> Bye,
> Roland
>
David Korn
[email protected]
_______________________________________________
ast-users mailing list
[email protected]
https://mailman.research.att.com/mailman/listinfo/ast-users
