: update leapseconds

  From 06b181c1370b368fc49ef3aede7e8ac7d47798c3 Mon Sep 17 00:00:00 2001
From: Gianni Tedesco <gianni@scaramanga.co.uk>
Date: Sun, 18 Mar 2018 05:21:56 +0900
Subject: [PATCH] Update leapseconds file per IERS Bulletein C 55

---
 timezone/leapseconds | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)
  

As I understand it, the leapseconds file (like most of the other files in the 
glibc timezone directory) is intended only for test data, and is not kept 
up-to-date. Perhaps this should be stated more clearly in the files' headers.

Another possibility would be to remove these files, as purposely out-of-date 
files cause confusion among glibc contributors.
  

On Sun, 2018-03-18 at 16:59 -0700, Paul Eggert wrote:
> As I understand it, the leapseconds file (like most of the other
> files in the 
> glibc timezone directory) is intended only for test data, and is not
> kept 
> up-to-date. Perhaps this should be stated more clearly in the files'
> headers.
> 
> Another possibility would be to remove these files, as purposely out-
> of-date 
> files cause confusion among glibc contributors.

Ah, I was under the impression they were used when converting in and
out of "right" timezones. But now I see that information is in the
tzinfo file itself.

My bad.

Gianni
  

On Sun, 18 Mar 2018, Paul Eggert wrote:

> As I understand it, the leapseconds file (like most of the other files in the
> glibc timezone directory) is intended only for test data, and is not kept
> up-to-date. Perhaps this should be stated more clearly in the files' headers.
> 
> Another possibility would be to remove these files, as purposely out-of-date
> files cause confusion among glibc contributors.

Maybe move them to e.g. timezone/test-sources?  (We already have 
timezone/testdata for *binary* timezone files used as test inputs to the 
timezone code, so it's probably a good idea to keep the different kinds of 
test inputs separate.)
  

On Mon, 2018-03-19 at 17:00 +0000, Joseph Myers wrote:
> On Sun, 18 Mar 2018, Paul Eggert wrote:
> 
> > As I understand it, the leapseconds file (like most of the other
> > files in the
> > glibc timezone directory) is intended only for test data, and is
> > not kept
> > up-to-date. Perhaps this should be stated more clearly in the
> > files' headers.
> > 
> > Another possibility would be to remove these files, as purposely
> > out-of-date
> > files cause confusion among glibc contributors.
> 
> Maybe move them to e.g. timezone/test-sources?  (We already have 
> timezone/testdata for *binary* timezone files used as test inputs to
> the 
> timezone code, so it's probably a good idea to keep the different
> kinds of 
> test inputs separate.)

Yeah, well, it's also getting shipped as part of the binary
distribution, at least on ubuntu, so if it's just test data then that's
weird...

Gianni
  

On Tue, 20 Mar 2018, Gianni Tedesco wrote:

> Yeah, well, it's also getting shipped as part of the binary
> distribution, at least on ubuntu, so if it's just test data then that's
> weird...

Distributions should be building from the separate tzdata package, not 
from glibc sources, which haven't installed the binary timezone data for 
several years.