Re: Seeking clarifaction of tai64nlocal and leap seconds

--20cf303b413b6613f504ca131876
Content-Type: text/plain; charset=ISO-8859-1

Brian,

Leap-seconds should really only matter when converting from a "pure" TAI64,
TAI64N, or TAI64NA timestamp (which simply counts seconds and fractions
thereof and does not care about hours, days, years, etc.) into a
human-readable format.

This is what the utility "tai64nlocal" does, for example - I use this tool
to view the contents of my multilog-generated log files. But multilog
itself doesn't need to care about calendrical hacks at all, and neither
does the upstream application.

Cheers,

    Peter

On Tue, Sep 18, 2012 at 3:08 PM, Brian Reichert <[email protected]>wrote:

> On Fri, Apr 01, 2011 at 02:47:41PM -0400, Brian Reichert wrote:
>
> In other threads, on other forums, I've asked about
>
> If libtai's leap-second table was up-to-date:
>
>   http://comments.gmane.org/gmane.comp.lib.libtai/361
>
> And here, about the relationship of tai64nlocal and leap seconds
>
>   http://comments.gmane.org/gmane.comp.djb.syslog/1319
>
> But I'm only now pawing through daemontools source:
>
>   http://cr.yp.to/daemontools/install.html
>
> And to my naive eye, it looks like multilog doesn't use libtai, and
> a such, I wonder how (if at all) it has a sense of leap seconds
> when it generates timestamps.
>
> The webpage/manpage for multilog says
>
>      Timestamping
>      The action
>
>        t
>
>      inserts an @, a precise timestamp, and a space in front of each line,
>      using the same format as tai64n. This is required to be the first
>      action.
>
> This only describes the format of the timestamp, and what isn't
> clear to me is if it _is_ a TAI timestamp.  (I, for years, have
> assumed that daemontool used libtai, and it's copy of leapsecs.dat.)
>
> Am I missing something?  What in the toolchain of mulitlog | tai64n
> 'knows' about leap seconds?
>
> --
> Brian Reichert                          <[email protected]>
> BSD admin/developer at large
>



-- 
Peter Wolfenden
650-801-7714

--20cf303b413b6613f504ca131876
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Brian,<div><br></div><div>Leap-seconds should really only matter when conve=
rting from a &quot;pure&quot; TAI64, TAI64N, or TAI64NA timestamp (which si=
mply counts seconds and fractions thereof and does not care about hours, da=
ys, years, etc.) into a human-readable format.</div>
<div><br></div><div>This is what the=A0utility &quot;tai64nlocal&quot; does=
, for example - I use this tool to view the contents of my multilog-generat=
ed log files. But multilog itself doesn&#39;t need to care about calendrica=
l hacks at all, and neither does the upstream application.</div>
<div><br></div><div>Cheers,</div><div><br></div><div>=A0 =A0 Peter</div><di=
v><br><div class=3D"gmail_quote">On Tue, Sep 18, 2012 at 3:08 PM, Brian Rei=
chert <span dir=3D"ltr">&lt;<a href=3D"mailto:[email protected]" target=
=3D"_blank">[email protected]</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">On Fri, Apr 01, 2011 at 02:47:41PM -0400, Br=
ian Reichert wrote:<br>
<br>
In other threads, on other forums, I&#39;ve asked about<br>
<br>
If libtai&#39;s leap-second table was up-to-date:<br>
<br>
=A0 <a href=3D"http://comments.gmane.org/gmane.comp.lib.libtai/361" target=
=3D"_blank">http://comments.gmane.org/gmane.comp.lib.libtai/361</a><br>
<br>
And here, about the relationship of tai64nlocal and leap seconds<br>
<br>
=A0 <a href=3D"http://comments.gmane.org/gmane.comp.djb.syslog/1319" target=
=3D"_blank">http://comments.gmane.org/gmane.comp.djb.syslog/1319</a><br>
<br>
But I&#39;m only now pawing through daemontools source:<br>
<br>
=A0 <a href=3D"http://cr.yp.to/daemontools/install.html" target=3D"_blank">=
http://cr.yp.to/daemontools/install.html</a><br>
<br>
And to my naive eye, it looks like multilog doesn&#39;t use libtai, and<br>
a such, I wonder how (if at all) it has a sense of leap seconds<br>
when it generates timestamps.<br>
<br>
The webpage/manpage for multilog says<br>
<br>
=A0 =A0 =A0Timestamping<br>
=A0 =A0 =A0The action<br>
<br>
=A0 =A0 =A0 =A0t<br>
<br>
=A0 =A0 =A0inserts an @, a precise timestamp, and a space in front of each =
line,<br>
=A0 =A0 =A0using the same format as tai64n. This is required to be the firs=
t<br>
=A0 =A0 =A0action.<br>
<br>
This only describes the format of the timestamp, and what isn&#39;t<br>
clear to me is if it _is_ a TAI timestamp. =A0(I, for years, have<br>
assumed that daemontool used libtai, and it&#39;s copy of leapsecs.dat.)<br=
>
<br>
Am I missing something? =A0What in the toolchain of mulitlog | tai64n<br>
&#39;knows&#39; about leap seconds?<br>
<div class=3D"HOEnZb"><div class=3D"h5"><br>
--<br>
Brian Reichert =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0&lt;<a hr=
ef=3D"mailto:[email protected]">[email protected]</a>&gt;<br>
BSD admin/developer at large<br>
</div></div></blockquote></div><br><br clear=3D"all"><div><br></div>-- <br>=
Peter Wolfenden<br>650-801-7714<br><br>
</div>

--20cf303b413b6613f504ca131876--