Re: [PATCH] t9604: Fix test for musl libc and new Debian

From: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Date: Apr 7, 2024, 01:45 UTC
Message-ID: <ZhH6wq5eTUTZS_zE@danh.dev>
In-Reply-To: <20240407013312.GD1085004@coredump.intra.peff.net>

On 2024-04-06 21:33:12-0400, Jeff King <peff@peff.net> wrote:

Show 22 quoted lines

> On Sat, Apr 06, 2024 at 10:29:10AM +0700, Đoàn Trần Công Danh wrote:
> 
> > CST6CDT and the like are POSIX timezone, with no rule for transition.
> > And POSIX doesn't enforce how to interpret the rule if it's omited.
> > Some libc resorted back to IANA (formerly Olson) db rules for those
> > timezones.  Other libc (e.g. musl) interpret that as no transition at
> > all [1].
> > 
> > In addition, distributions (notoriously Debian-derived, which uses IANA
> > db for CST6CDT and the like) started to split "legacy" timezones
> > like CST6CDT, EST5EDT into `tzdata-legacy', which will not be installed
> > by default [2].
> > 
> > In those cases, t9604 will run into failure.
> > 
> > Let's switch to POSIX timezone with rules to change timezone.
> 
> This made me wonder if we are losing EST5, etc. We use that in t0006,
> for example. But I guess not, since I do not have tzdata-legacy
> installed (I am on Debian unstable) and haven't run into issues (I
> didn't notice the cvsimport one because I lack other prereqs to run
> those tests).

Nah, EST5 is a conformance POSIX timezone.  It read:

    The timezone name is EST, offset is 5hours behinds Universal timezone.

You can check by trying today (or on anyday with DST on):

	TZ=EST5 date
	TZ=EST5EDT date
	TZ=America/New_York date

- The first one will always interpret 5 hours behinds UTC.
- The second one is implementation defined behavior, on glibc system,
  it will depends on the existence of /usr/share/zoneinfo/EST5EDT
- The third one will interpret today time as 4 hours behinds UTC.

glibc normally check if a timezone exist in /usr/share/zoneinfo first, if not, it will interpret by POSIX rule.

-- 
Danh