`time` doesn't say seconds

`time` doesn't say seconds

[Ram Rachum]

[-- Attachment #1: Type: text/plain, Size: 349 bytes --]

Observe this call to `time`:

```
(jax_env) ➜  purejaxrl git:(main) ✗ time python foo.py
python foo.py  11.41s user 0.90s system 109% cpu 11.230 total
```

If I understand correctly, the 11.230 number is 11.230 seconds, but for
some reason it doesn't have "s" after it like 11.41s and 0.90s. I think
that it should.


Thanks,
Ram.

[-- Attachment #2: Type: text/html, Size: 487 bytes --]

Re: `time` doesn't say seconds

[Bart Schaefer]

On Fri, Nov 10, 2023 at 5:25 AM Ram Rachum <ram@rachum.com> wrote:
>
> If I understand correctly, the 11.230 number is 11.230 seconds, but for some reason it doesn't have "s" after it like 11.41s and 0.90s. I think that it should.

TIMEFMT='%J %U user %S system %P cpu %E total'

The default format instead uses %*E which drops the trailing "s" and
uses HH:MM:SS.TTT instead, but the HH and MM are also dropped when
zero.

Re: `time` doesn't say seconds

[Ram Rachum]

[-- Attachment #1: Type: text/plain, Size: 645 bytes --]

On Fri, Nov 10, 2023, 19:26 Bart Schaefer <schaefer@brasslantern.com> wrote:

> On Fri, Nov 10, 2023 at 5:25 AM Ram Rachum <ram@rachum.com> wrote:
> >
> > If I understand correctly, the 11.230 number is 11.230 seconds, but for
> some reason it doesn't have "s" after it like 11.41s and 0.90s. I think
> that it should.
>
> TIMEFMT='%J %U user %S system %P cpu %E total'
>
> The default format instead uses %*E which drops the trailing "s" and
> uses HH:MM:SS.TTT instead, but the HH and MM are also dropped when
> zero.
>


Thanks Bart. Given this explanation, what do you think should be done here,
if anything at all?

>

[-- Attachment #2: Type: text/html, Size: 1283 bytes --]

Re: `time` doesn't say seconds

[Bart Schaefer]

On Fri, Nov 10, 2023 at 9:28 AM Ram Rachum <ram@rachum.com> wrote:
>
> Thanks Bart. Given this explanation, what do you think should be done here, if anything at all

I think you should assign TIMEFMT in your .z* files somewhere.  The
default is documented.

Re: `time` doesn't say seconds

[Peter Stephenson]

[-- Attachment #1: Type: text/plain, Size: 1338 bytes --]

> On 10/11/2023 17:26 GMT Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Fri, Nov 10, 2023 at 5:25 AM Ram Rachum <ram@rachum.com> wrote:
> >
> > If I understand correctly, the 11.230 number is 11.230 seconds, but
> > for some reason it doesn't have "s" after it like 11.41s and
> > 0.90s. I think that it should.
>
> TIMEFMT='%J %U user %S system %P cpu %E total'
>
> The default format instead uses %*E which drops the trailing "s" and
> uses HH:MM:SS.TTT instead, but the HH and MM are also dropped when
> zero.

I meant to follow up to this one.

It could be made more consistent by putting back the s in that last
case, though.  Without the colon indicating the meaning of the numbers
(though MM:SS and HH:MM are also a bit ambiguous) this probably makes
more sense.

I don't know if anyone is likely to be automatically matching the
current format?  Given the variability of this element presumably not.

diff --git a/Src/jobs.c b/Src/jobs.c
index a3b9f667a..bb9246837 100644
--- a/Src/jobs.c
+++ b/Src/jobs.c
@@ -730,7 +730,7 @@ printhhmmss(double secs)
     else if (mins)
        fprintf(stderr,      "%d:%05.2f",        mins, secs);
     else
-       fprintf(stderr,           "%.3f",              secs);
+       fprintf(stderr,           "%.3fs",              secs);
 }
 
 static void

[-- Attachment #2: Type: text/html, Size: 1618 bytes --]

Re: `time` doesn't say seconds

[Bart Schaefer]

On Tue, Nov 21, 2023 at 1:25 AM Peter Stephenson
<p.w.stephenson@ntlworld.com> wrote:
>
> I don't know if anyone is likely to be automatically matching the
> current format?  Given the variability of this element presumably not.

The doc should be updated, though.  It specifies the default without
the "s".  Your patch also alters the definition of %*E in the value of
TIMEFMT.

Re: `time` doesn't say seconds

[Peter Stephenson]

> On 22/11/2023 03:36 GMT Bart Schaefer <schaefer@brasslantern.com> wrote:
> On Tue, Nov 21, 2023 at 1:25 AM Peter Stephenson
> <p.w.stephenson@ntlworld.com> wrote:
> >
> > I don't know if anyone is likely to be automatically matching the
> > current format?  Given the variability of this element presumably not.
> 
> The doc should be updated, though.  It specifies the default without
> the "s".  Your patch also alters the definition of %*E in the value of
> TIMEFMT.

This is the only place I can see with late level of detail.

pws

--- a/Doc/Zsh/params.yo
+++ b/Doc/Zsh/params.yo
@@ -1721,7 +1721,8 @@ endsitem()
 A star may be inserted between the percent sign and flags printing time
 (e.g., `tt(%*E)'); this causes the time to be printed in
 `var(hh)tt(:)var(mm)tt(:)var(ss)tt(.)var(ttt)'
-format (hours and minutes are only printed if they are not zero).
+format.  Hours and minutes are only printed if they are not zero;
+an `s' is appended if they are both zero.
 Alternatively, `tt(m)' or `tt(u)' may be used (e.g., `tt(%mE)') to produce
 time output in milliseconds or microseconds, respectively.
 )
diff --git a/Src/jobs.c b/Src/jobs.c
index a3b9f667a..bb9246837 100644
--- a/Src/jobs.c
+++ b/Src/jobs.c
@@ -730,7 +730,7 @@ printhhmmss(double secs)
     else if (mins)
 	fprintf(stderr,      "%d:%05.2f",        mins, secs);
     else
-	fprintf(stderr,           "%.3f",              secs);
+	fprintf(stderr,           "%.3fs",              secs);
 }
 
 static void

Re: `time` doesn't say seconds

[Mark J. Reed]

[-- Attachment #1: Type: text/plain, Size: 768 bytes --]

Sorry if I'm missing some vital context (I here am new), but if the goal is
to include the "s" units in the default output of time, that's just a
matter of changing `%*E` to `%E` in the default value of TIMEFMT, right?

On Tue, Nov 21, 2023 at 10:36 PM Bart Schaefer <schaefer@brasslantern.com>
wrote:

> On Tue, Nov 21, 2023 at 1:25 AM Peter Stephenson
> <p.w.stephenson@ntlworld.com> wrote:
> >
> > I don't know if anyone is likely to be automatically matching the
> > current format?  Given the variability of this element presumably not.
>
> The doc should be updated, though.  It specifies the default without
> the "s".  Your patch also alters the definition of %*E in the value of
> TIMEFMT.
>
>

-- 
Mark J. Reed <markjreed@gmail.com>

[-- Attachment #2: Type: text/html, Size: 1367 bytes --]

Re: `time` doesn't say seconds

[Bart Schaefer]

On Wed, Nov 22, 2023 at 12:23 PM Mark J. Reed <markjreed@gmail.com> wrote:
>
> Sorry if I'm missing some vital context (I here am new), but if the goal is to include the "s" units in the default output of time, that's just a matter of changing `%*E` to `%E` in the default value of TIMEFMT, right?

%E is always in seconds, so something that ran for an hour-ish would
say something like 3612.03s instead of e.g. 01:00:12.0297 for %*E