i:
The compile-time flag NOSOLAR has been removed, as nowadays the
benefit of slightly shrinking runtime table size is outweighed by the
cost of disallowing potential future updates that exceed old limits.
h:
Fix localtime overflow bugs with 32-bit unsigned time_t.
zdump no longer assumes sscanf returns maximal values on overflow.
g:
'zic' now runs on platforms that lack both hard links and symlinks.
(Thanks to Theo Veenker for reporting the problem, for MinGW.)
Also, fix some bugs on platforms that lack hard links but have symlinks.
'zic -v' again warns that Asia/Tehran has no POSIX environment variable
to predict the far future, fixing a bug introduced in 2013e.
f:
The types of the global variables 'timezone' and 'altzone' (if present)
have been changed back to 'long'. This is required for 'timezone'
by POSIX, and for 'altzone' by common practice, e.g., Solaris 11.
These variables were originally 'long' in the tz code, but were
mistakenly changed to 'time_t' in 1987; nobody reported the
incompatibility until now. The difference matters on x32, where
'long' is 32 bits and 'time_t' is 64. (Thanks to Elliott Hughes.)
Changes affecting API
The 'zic' command now outputs a dummy transition when far-future
data can't be summarized using a TZ string, and uses a 402-year
window rather than a 400-year window. For the current data, this
affects only the Asia/Tehran file. It does not affect any of the
time stamps that this file represents, so zdump outputs the same
information as before. (Thanks to Andrew Main (Zefram).)
The 'date' command has a new '-r' option, which lets you specify
the integer time to display, a la FreeBSD.
The 'tzselect' command has two new options '-c' and '-n', which lets you
select a zone based on latitude and longitude.
The 'zic' command's '-v' option now warns about constructs that
require the new version-3 binary file format. (Thanks to Arthur
David Olson for the suggestion.)
Support for floating-point time_t has been removed.
It was always dicey, and POSIX no longer requires it.
(Thanks to Eric Blake for suggesting to the POSIX committee to
remove it, and thanks to Alan Barrett, Clive D.W. Feather, Andy
Heninger, Arthur David Olson, and Alois Treindl, for reporting
bugs and elucidating some of the corners of the old floating-point
implementation.)
The signatures of 'offtime', 'timeoff', and 'gtime' have been
changed back to the old practice of using 'long' to represent UT
offsets. This had been inadvertently and mistakenly changed to
'int_fast32_t'. (Thanks to Christos Zoulos.)
The code avoids undefined behavior on integer overflow in some
more places, including gmtime, localtime, mktime and zdump.
Changes affecting the zdump utility
zdump now outputs "UT" when referring to Universal Time, not "UTC".
"UTC" does not make sense for time stamps that predate the introduction
of UTC, whereas "UT", a more-generic term, does. (Thanks to Steve Allen
for clarifying UT vs UTC.)
Data changes affecting behavior of tzselect and similar programs
Country code BQ is now called the more-common name "Caribbean Netherlands"
rather than the more-official "Bonaire, St Eustatius & Saba".
Remove from zone.tab the names America/Montreal, America/Shiprock,
and Antarctica/South_Pole, as they are equivalent to existing
same-country-code zones for post-1970 time stamps. The data for
these names are unchanged, so the names continue to work as before.
Changes affecting code internals
zic -c now runs way faster on 64-bit hosts when given large numbers.
zic now uses vfprintf to avoid allocating and freeing some memory.
tzselect now computes the list of continents from the data,
rather than have it hard-coded.
Minor changes pacify GCC 4.7.3 and GCC 4.8.1.
Changes affecting the build procedure
The 'leapseconds' file is now generated automatically from a
new file 'leap-seconds.list', which is a copy of
<ftp://time.nist.gov/pub/leap-seconds.list>.
A new source file 'leapseconds.awk' implements this.
The goal is simplification of the future maintenance of 'leapseconds'.
When building the 'posix' or 'right' subdirectories, if the
subdirectory would be a copy of the default subdirectory, it is
now made a symbolic link if that is supported. This saves about
2 MB of file system space.
The links America/Shiprock and Antarctica/South_Pole have been
moved to the 'backward' file. This affects only nondefault builds
that omit 'backward'.
Changes affecting documentation and commentary
Changes to the 'tzfile' man page
It now mentions that the binary file format may be extended in
future versions by appending data.
It now refers to the 'zdump' and 'zic' man pages.
Changes to the 'zic' man page
It lists conditions that elicit a warning with '-v'.
It says that the behavior is unspecified when duplicate names
are given, or if the source of one link is the target of another.
Its examples are updated to match the latest data.
The definition of white space has been clarified slightly.
(Thanks to Michael Deckers.)
Changes to the 'Theory' file
There is a new section about the accuracy of the tz database,
describing the many ways that errors can creep in, and
explaining why so many of the pre-1970 time stamps are wrong or
misleading (thanks to Steve Allen, Lester Caine, and Garrett
Wollman for discussions that contributed to this).
The 'Theory' file describes LMT better (this follows a
suggestion by Guy Harris).
It refers to the 2013 edition of POSIX rather than the 2004 edition.
It's mentioned that excluding 'backward' should not affect the
other data, and it suggests at least one zone.tab name per
inhabited country (thanks to Stephen Colebourne).
Some longstanding restrictions on names are documented, e.g.,
'America/New_York' precludes 'America/New_York/Bronx'.
It gives more reasons for the 1970 cutoff.
It now mentions which time_t variants are supported, such as
signed integer time_t. (Thanks to Paul Goyette for reporting
typos in an experimental version of this change.)
(Thanks to Philip Newton for correcting typos in these changes.)
Documentation and commentary is more careful to distinguish UT in
general from UTC in particular. (Thanks to Steve Allen.)
Add a better source for the Zurich 1894 transition.
(Thanks to Pierre-Yves Berger.)
Update shapefile citations in tz-link.htm. (Thanks to Guy Harris.)
perform the conversion, but returns EINVAL when the time requested would fall
in the DST gap, or is not representable in the timezone requested, and document
this.
don't require locking and can operate on user-specified timezones
as opposed to having to alter the environment to change a timezone.
This work was presented to the tzcode folks and it was generally
accepted, but there seems to be a lot of inertia.
- asctime{,_r}, ctime{,_r} may return NULL; document that, and avoid coredumps.
- gmtime{,_r}, localtime{,_r} may return NULL and set EOVERFLOW, document and
set errno.
- when mktime returns (time_t)-1, make it set EOVERFLOW and document it.
XXX: Should be pulled up to 5.x
- now understands 64bit time_t and 64bit data in timezone files.
- localtime(), gmtime(), asctime() and ctime() may now fail with
a NULL result if time_t cannot be represented by struct tm.
wiz